OOM killer doesn't work properly, leads to a frozen OS
Clash Royale CLAN TAG#URR8PPP
up vote
10
down vote
favorite
Since years, the OOM killer of my operating system doesn't work properly and leads to a frozen system.
When the memory usage is very high, the whole system tends to freeze for hours or even days (the maximum that I have recorded is 7 days before a reset), instead of killing processes to free the memory.
In this situation, the iowait is very very high (~ 70%).
The tool: iotop has showed that every programs are reading at a very high throughput (per tens of MB/sec) from my hard drive.
What those programs are reading ?
- The directory hierarchy ?
- The executable code itself ?
I don't exactly now.
I use an uptodate ArchLinux (currently: 4.9.27-1-lts), have 16 GB of physical ram, and no enabled swap partition.
Because of the amount of ram that I have, I don't want to enable a swap partition, since it would just delay the apparition of the issue.
To me the problem is caused by the fact that Linux drops essential data from the caches, which leads to a frozen system because it has to read everything, every time from the hard drive.
I even wonder if Linux wouldn't drop the executable code pages of running programs, which would explain why programs that normally don't read a lot of data, behave this way in this situation.
I have tried several things in the hope to fix this issue.
One was to set /proc/sys/vm/min_free_kbytes to 1000000 (1 GB).
Because this 1 GB should remains free, I thought that this memory would be reserved by Linux to cache important data.
But it hasn't worked.
Also, I think useful to add that even if it sounds great in theory, restricting the size of the virtual memory to the size of the physical memory, by defining /proc/sys/vm/overcommit_memory to 2 isn't decently technically possible in my situation, because the kind of applications that I use, require more virtual memory than they effectively use for some reasons.
According to the file /proc/meminfo, the Commited_AS value is often higher than the double of the physical ram on my system (16 GB, Commited_AS is often > 32 GB).
I have experienced this problem with /proc/sys/vm/overcommit_memory to its default value: 0, and since a while I have defined it to: 1, because I prefer programs to be killed by the OOM killer rather than behaving wrongly because they don't check the return values of malloc, when the allocations are refused.
When I was talking about this issue on IRC, I have met other Linux users who have experienced this very same problem, so I guess that a lot of users are concerned by this.
To me this is not acceptable since even Windows deals better with high memory usage.
If you need more information, have a suggestion, please tell me.
linux arch-linux out-of-memory
asked Jun 25 '17 at 19:45
parasite
1245
 |Â
show 6 more comments
up vote
10
down vote
favorite
Since years, the OOM killer of my operating system doesn't work properly and leads to a frozen system.
When the memory usage is very high, the whole system tends to freeze for hours or even days (the maximum that I have recorded is 7 days before a reset), instead of killing processes to free the memory.
In this situation, the iowait is very very high (~ 70%).
The tool: iotop has showed that every programs are reading at a very high throughput (per tens of MB/sec) from my hard drive.
What those programs are reading ?
- The directory hierarchy ?
- The executable code itself ?
I don't exactly now.
I use an uptodate ArchLinux (currently: 4.9.27-1-lts), have 16 GB of physical ram, and no enabled swap partition.
Because of the amount of ram that I have, I don't want to enable a swap partition, since it would just delay the apparition of the issue.
To me the problem is caused by the fact that Linux drops essential data from the caches, which leads to a frozen system because it has to read everything, every time from the hard drive.
I even wonder if Linux wouldn't drop the executable code pages of running programs, which would explain why programs that normally don't read a lot of data, behave this way in this situation.
I have tried several things in the hope to fix this issue.
One was to set /proc/sys/vm/min_free_kbytes to 1000000 (1 GB).
Because this 1 GB should remains free, I thought that this memory would be reserved by Linux to cache important data.
But it hasn't worked.
Also, I think useful to add that even if it sounds great in theory, restricting the size of the virtual memory to the size of the physical memory, by defining /proc/sys/vm/overcommit_memory to 2 isn't decently technically possible in my situation, because the kind of applications that I use, require more virtual memory than they effectively use for some reasons.
According to the file /proc/meminfo, the Commited_AS value is often higher than the double of the physical ram on my system (16 GB, Commited_AS is often > 32 GB).
I have experienced this problem with /proc/sys/vm/overcommit_memory to its default value: 0, and since a while I have defined it to: 1, because I prefer programs to be killed by the OOM killer rather than behaving wrongly because they don't check the return values of malloc, when the allocations are refused.
When I was talking about this issue on IRC, I have met other Linux users who have experienced this very same problem, so I guess that a lot of users are concerned by this.
To me this is not acceptable since even Windows deals better with high memory usage.
If you need more information, have a suggestion, please tell me.
linux arch-linux out-of-memory
asked Jun 25 '17 at 19:45
parasite
1245
1
I think this is what you should expect, if you're thrashing, but you're not really approaching 100% "used" i.e. there is too much memory usage which is file-backed, counted as "buff/cache". (Ugh, this phrasing assumes your tmpfs allocations are trivial, as these show up as "buff/cache", but cannot be paged out to a physical filesystem).min_free_kbytesis not relevant, it's not a reserve for cached pages. AFAICT none of the vm sysctls allow reserving any memory specifically for cached pages, i.e. limiting MAP_ANONYMOUS allocations :(.
– sourcejedi
Jun 28 '17 at 19:41
Thank you for having confirmed that setting another value to:min_free_kbytesis not the right way to solve this issue. However, I disagree with your statement: "this is what you should expect". What I do expect is a system good enough to free the memory, by killing non vital processes, in high memory usage, and certainly not a frozen system, unusable for days.
– parasite
Jun 29 '17 at 14:53
Ugh again. I was trying to make a technical statement, not a normative one. To be clear, I agree it's annoying there seems no way to reserve memory for cached pages, which you point out is the obvious way to try and mitigate this. I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process
– sourcejedi
Jun 29 '17 at 15:55
I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process. I imagine the latter is easier... there's less ambiguity about whether the app is expected to use the page again later. Running out of memory sucks and Linux hasn't managed to mitigate that very much.
– sourcejedi
Jun 29 '17 at 16:01
1
I've been looking for a solution for this exact issue for years now without any success. I believe I first noticed the problem after replacing my HDD by an SSD, which also entailed me disabling swapping altogether, but I can't really guarantee that it never happened before these changes, so it might be unrelated. I'm on Archlinux btw.
– brunocodutra
Aug 8 '17 at 11:59
 |Â
show 6 more comments
up vote
10
down vote
favorite
up vote
10
down vote
favorite
Since years, the OOM killer of my operating system doesn't work properly and leads to a frozen system.
When the memory usage is very high, the whole system tends to freeze for hours or even days (the maximum that I have recorded is 7 days before a reset), instead of killing processes to free the memory.
In this situation, the iowait is very very high (~ 70%).
The tool: iotop has showed that every programs are reading at a very high throughput (per tens of MB/sec) from my hard drive.
What those programs are reading ?
- The directory hierarchy ?
- The executable code itself ?
I don't exactly now.
I use an uptodate ArchLinux (currently: 4.9.27-1-lts), have 16 GB of physical ram, and no enabled swap partition.
Because of the amount of ram that I have, I don't want to enable a swap partition, since it would just delay the apparition of the issue.
To me the problem is caused by the fact that Linux drops essential data from the caches, which leads to a frozen system because it has to read everything, every time from the hard drive.
I even wonder if Linux wouldn't drop the executable code pages of running programs, which would explain why programs that normally don't read a lot of data, behave this way in this situation.
I have tried several things in the hope to fix this issue.
One was to set /proc/sys/vm/min_free_kbytes to 1000000 (1 GB).
Because this 1 GB should remains free, I thought that this memory would be reserved by Linux to cache important data.
But it hasn't worked.
Also, I think useful to add that even if it sounds great in theory, restricting the size of the virtual memory to the size of the physical memory, by defining /proc/sys/vm/overcommit_memory to 2 isn't decently technically possible in my situation, because the kind of applications that I use, require more virtual memory than they effectively use for some reasons.
According to the file /proc/meminfo, the Commited_AS value is often higher than the double of the physical ram on my system (16 GB, Commited_AS is often > 32 GB).
I have experienced this problem with /proc/sys/vm/overcommit_memory to its default value: 0, and since a while I have defined it to: 1, because I prefer programs to be killed by the OOM killer rather than behaving wrongly because they don't check the return values of malloc, when the allocations are refused.
When I was talking about this issue on IRC, I have met other Linux users who have experienced this very same problem, so I guess that a lot of users are concerned by this.
To me this is not acceptable since even Windows deals better with high memory usage.
If you need more information, have a suggestion, please tell me.
linux arch-linux out-of-memory
asked Jun 25 '17 at 19:45
parasite
1245
Since years, the OOM killer of my operating system doesn't work properly and leads to a frozen system.
When the memory usage is very high, the whole system tends to freeze for hours or even days (the maximum that I have recorded is 7 days before a reset), instead of killing processes to free the memory.
In this situation, the iowait is very very high (~ 70%).
The tool: iotop has showed that every programs are reading at a very high throughput (per tens of MB/sec) from my hard drive.
What those programs are reading ?
- The directory hierarchy ?
- The executable code itself ?
I don't exactly now.
I use an uptodate ArchLinux (currently: 4.9.27-1-lts), have 16 GB of physical ram, and no enabled swap partition.
Because of the amount of ram that I have, I don't want to enable a swap partition, since it would just delay the apparition of the issue.
To me the problem is caused by the fact that Linux drops essential data from the caches, which leads to a frozen system because it has to read everything, every time from the hard drive.
I even wonder if Linux wouldn't drop the executable code pages of running programs, which would explain why programs that normally don't read a lot of data, behave this way in this situation.
I have tried several things in the hope to fix this issue.
One was to set /proc/sys/vm/min_free_kbytes to 1000000 (1 GB).
Because this 1 GB should remains free, I thought that this memory would be reserved by Linux to cache important data.
But it hasn't worked.
Also, I think useful to add that even if it sounds great in theory, restricting the size of the virtual memory to the size of the physical memory, by defining /proc/sys/vm/overcommit_memory to 2 isn't decently technically possible in my situation, because the kind of applications that I use, require more virtual memory than they effectively use for some reasons.
According to the file /proc/meminfo, the Commited_AS value is often higher than the double of the physical ram on my system (16 GB, Commited_AS is often > 32 GB).
I have experienced this problem with /proc/sys/vm/overcommit_memory to its default value: 0, and since a while I have defined it to: 1, because I prefer programs to be killed by the OOM killer rather than behaving wrongly because they don't check the return values of malloc, when the allocations are refused.
When I was talking about this issue on IRC, I have met other Linux users who have experienced this very same problem, so I guess that a lot of users are concerned by this.
To me this is not acceptable since even Windows deals better with high memory usage.
If you need more information, have a suggestion, please tell me.
linux arch-linux out-of-memory
linux arch-linux out-of-memory
asked Jun 25 '17 at 19:45
parasite
1245
asked Jun 25 '17 at 19:45
parasite
1245
edited Jul 3 '17 at 12:55
asked Jun 25 '17 at 19:45
parasite
1245
asked Jun 25 '17 at 19:45
parasite
1245
asked Jun 25 '17 at 19:45
parasite
1245
1245
1
I think this is what you should expect, if you're thrashing, but you're not really approaching 100% "used" i.e. there is too much memory usage which is file-backed, counted as "buff/cache". (Ugh, this phrasing assumes your tmpfs allocations are trivial, as these show up as "buff/cache", but cannot be paged out to a physical filesystem).min_free_kbytesis not relevant, it's not a reserve for cached pages. AFAICT none of the vm sysctls allow reserving any memory specifically for cached pages, i.e. limiting MAP_ANONYMOUS allocations :(.
– sourcejedi
Jun 28 '17 at 19:41
Thank you for having confirmed that setting another value to:min_free_kbytesis not the right way to solve this issue. However, I disagree with your statement: "this is what you should expect". What I do expect is a system good enough to free the memory, by killing non vital processes, in high memory usage, and certainly not a frozen system, unusable for days.
– parasite
Jun 29 '17 at 14:53
Ugh again. I was trying to make a technical statement, not a normative one. To be clear, I agree it's annoying there seems no way to reserve memory for cached pages, which you point out is the obvious way to try and mitigate this. I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process
– sourcejedi
Jun 29 '17 at 15:55
I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process. I imagine the latter is easier... there's less ambiguity about whether the app is expected to use the page again later. Running out of memory sucks and Linux hasn't managed to mitigate that very much.
– sourcejedi
Jun 29 '17 at 16:01
1
I've been looking for a solution for this exact issue for years now without any success. I believe I first noticed the problem after replacing my HDD by an SSD, which also entailed me disabling swapping altogether, but I can't really guarantee that it never happened before these changes, so it might be unrelated. I'm on Archlinux btw.
– brunocodutra
Aug 8 '17 at 11:59
 |Â
show 6 more comments
1
I think this is what you should expect, if you're thrashing, but you're not really approaching 100% "used" i.e. there is too much memory usage which is file-backed, counted as "buff/cache". (Ugh, this phrasing assumes your tmpfs allocations are trivial, as these show up as "buff/cache", but cannot be paged out to a physical filesystem).min_free_kbytesis not relevant, it's not a reserve for cached pages. AFAICT none of the vm sysctls allow reserving any memory specifically for cached pages, i.e. limiting MAP_ANONYMOUS allocations :(.
– sourcejedi
Jun 28 '17 at 19:41
Thank you for having confirmed that setting another value to:min_free_kbytesis not the right way to solve this issue. However, I disagree with your statement: "this is what you should expect". What I do expect is a system good enough to free the memory, by killing non vital processes, in high memory usage, and certainly not a frozen system, unusable for days.
– parasite
Jun 29 '17 at 14:53
Ugh again. I was trying to make a technical statement, not a normative one. To be clear, I agree it's annoying there seems no way to reserve memory for cached pages, which you point out is the obvious way to try and mitigate this. I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process
– sourcejedi
Jun 29 '17 at 15:55
I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process. I imagine the latter is easier... there's less ambiguity about whether the app is expected to use the page again later. Running out of memory sucks and Linux hasn't managed to mitigate that very much.
– sourcejedi
Jun 29 '17 at 16:01
1
I've been looking for a solution for this exact issue for years now without any success. I believe I first noticed the problem after replacing my HDD by an SSD, which also entailed me disabling swapping altogether, but I can't really guarantee that it never happened before these changes, so it might be unrelated. I'm on Archlinux btw.
– brunocodutra
Aug 8 '17 at 11:59
1
1
I think this is what you should expect, if you're thrashing, but you're not really approaching 100% "used" i.e. there is too much memory usage which is file-backed, counted as "buff/cache". (Ugh, this phrasing assumes your tmpfs allocations are trivial, as these show up as "buff/cache", but cannot be paged out to a physical filesystem).
min_free_kbytes is not relevant, it's not a reserve for cached pages. AFAICT none of the vm sysctls allow reserving any memory specifically for cached pages, i.e. limiting MAP_ANONYMOUS allocations :(.– sourcejedi
Jun 28 '17 at 19:41
I think this is what you should expect, if you're thrashing, but you're not really approaching 100% "used" i.e. there is too much memory usage which is file-backed, counted as "buff/cache". (Ugh, this phrasing assumes your tmpfs allocations are trivial, as these show up as "buff/cache", but cannot be paged out to a physical filesystem).
min_free_kbytes is not relevant, it's not a reserve for cached pages. AFAICT none of the vm sysctls allow reserving any memory specifically for cached pages, i.e. limiting MAP_ANONYMOUS allocations :(.– sourcejedi
Jun 28 '17 at 19:41
Thank you for having confirmed that setting another value to:
min_free_kbytes is not the right way to solve this issue. However, I disagree with your statement: "this is what you should expect". What I do expect is a system good enough to free the memory, by killing non vital processes, in high memory usage, and certainly not a frozen system, unusable for days.– parasite
Jun 29 '17 at 14:53
Thank you for having confirmed that setting another value to:
min_free_kbytes is not the right way to solve this issue. However, I disagree with your statement: "this is what you should expect". What I do expect is a system good enough to free the memory, by killing non vital processes, in high memory usage, and certainly not a frozen system, unusable for days.– parasite
Jun 29 '17 at 14:53
Ugh again. I was trying to make a technical statement, not a normative one. To be clear, I agree it's annoying there seems no way to reserve memory for cached pages, which you point out is the obvious way to try and mitigate this. I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process
– sourcejedi
Jun 29 '17 at 15:55
Ugh again. I was trying to make a technical statement, not a normative one. To be clear, I agree it's annoying there seems no way to reserve memory for cached pages, which you point out is the obvious way to try and mitigate this. I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process
– sourcejedi
Jun 29 '17 at 15:55
I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process. I imagine the latter is easier... there's less ambiguity about whether the app is expected to use the page again later. Running out of memory sucks and Linux hasn't managed to mitigate that very much.
– sourcejedi
Jun 29 '17 at 16:01
I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process. I imagine the latter is easier... there's less ambiguity about whether the app is expected to use the page again later. Running out of memory sucks and Linux hasn't managed to mitigate that very much.
– sourcejedi
Jun 29 '17 at 16:01
1
1
I've been looking for a solution for this exact issue for years now without any success. I believe I first noticed the problem after replacing my HDD by an SSD, which also entailed me disabling swapping altogether, but I can't really guarantee that it never happened before these changes, so it might be unrelated. I'm on Archlinux btw.
– brunocodutra
Aug 8 '17 at 11:59
I've been looking for a solution for this exact issue for years now without any success. I believe I first noticed the problem after replacing my HDD by an SSD, which also entailed me disabling swapping altogether, but I can't really guarantee that it never happened before these changes, so it might be unrelated. I'm on Archlinux btw.
– brunocodutra
Aug 8 '17 at 11:59
 |Â
show 6 more comments
2 Answers
2
active
oldest
votes
up vote
1
down vote
I beleve this was a kernel bug fixed sometime before 4.14.58 / 4.9.115;
I had similar issues close to the version you reported which are now fixed.
answered Jul 27 at 14:33
user1133275
2,277412
add a comment |Â
up vote
1
down vote
I've found two explanations(of the same thing) as to why kswapd0 does constant disk reading happens well before OOM-killer kills the offending process:
- see the answer and comment of this askubuntu SE answer
- see the answer and David Schwartz's comments of this answer on unix SE
I'll quote here the comment from 1. which really opened my eyes as to why I was getting constant disk reading while everything was frozen:
For example, consider a case where you have zero swap and system is
nearly running out of RAM. The kernel will take memory from e.g.
Firefox (it can do this because Firefox is running executable code
that has been loaded from disk - the code can be loaded from disk
again if needed). If Firefox then needs to access that RAM again N
seconds later, the CPU generates "hard fault" which forces Linux to
free some RAM (e.g. take some RAM from another process), load the
missing data from disk and then allow Firefox to continue as usual.
This is pretty similar to normal swapping and kswapd0 does it. – Mikko
Rantalainen Feb 15 at 13:08
If anyone has a way as to how to disable this behavior(maybe recompile kernel with what options?), please let me know as soon as possible! Much appreciated, thanks!EDIT: I've just found out and tested that vm.overcommit_memory=2 avoids the disk thrashing that leads to frozen OS (even though the OOM-killer doesn't get a change to trigger, which makes sense because it only triggered well after the disk-thrashing anyway, with vm.overcommit_memory=0) like: cc1plus: out of memory allocating 127440 bytes after a total of 897024 bytes
EDIT2 ok the above worked for me with the default vm.overcommit_ratio=50 but if I set it to 200 then the disk-thrashing is back!
EDIT3 ignore these 2 EDITs, they are not the way to fix this because processes that wouldn't have died before die now sooner, also vm.overcommit_ratio=0 will cause everything to die, or something. I'll try to find a better way, likely needing kernel recompile and I'm looking into the relevant GFP flags...
EDIT4: I found a way, through patching kernel, that works for me; see the patch inside this question.
answered Aug 19 at 10:23
Marcus Linsner
15914
add a comment |Â
2 Answers
2
active
oldest
votes
2 Answers
2
active
oldest
votes
active
oldest
votes
active
oldest
votes
up vote
1
down vote
I beleve this was a kernel bug fixed sometime before 4.14.58 / 4.9.115;
I had similar issues close to the version you reported which are now fixed.
answered Jul 27 at 14:33
user1133275
2,277412
add a comment |Â
up vote
1
down vote
I beleve this was a kernel bug fixed sometime before 4.14.58 / 4.9.115;
I had similar issues close to the version you reported which are now fixed.
answered Jul 27 at 14:33
user1133275
2,277412
add a comment |Â
up vote
1
down vote
up vote
1
down vote
I beleve this was a kernel bug fixed sometime before 4.14.58 / 4.9.115;
I had similar issues close to the version you reported which are now fixed.
answered Jul 27 at 14:33
user1133275
2,277412
I beleve this was a kernel bug fixed sometime before 4.14.58 / 4.9.115;
I had similar issues close to the version you reported which are now fixed.
answered Jul 27 at 14:33
user1133275
2,277412
answered Jul 27 at 14:33
user1133275
2,277412
answered Jul 27 at 14:33
user1133275
2,277412
answered Jul 27 at 14:33
user1133275
2,277412
2,277412
add a comment |Â
add a comment |Â
up vote
1
down vote
I've found two explanations(of the same thing) as to why kswapd0 does constant disk reading happens well before OOM-killer kills the offending process:
- see the answer and comment of this askubuntu SE answer
- see the answer and David Schwartz's comments of this answer on unix SE
I'll quote here the comment from 1. which really opened my eyes as to why I was getting constant disk reading while everything was frozen:
For example, consider a case where you have zero swap and system is
nearly running out of RAM. The kernel will take memory from e.g.
Firefox (it can do this because Firefox is running executable code
that has been loaded from disk - the code can be loaded from disk
again if needed). If Firefox then needs to access that RAM again N
seconds later, the CPU generates "hard fault" which forces Linux to
free some RAM (e.g. take some RAM from another process), load the
missing data from disk and then allow Firefox to continue as usual.
This is pretty similar to normal swapping and kswapd0 does it. – Mikko
Rantalainen Feb 15 at 13:08
If anyone has a way as to how to disable this behavior(maybe recompile kernel with what options?), please let me know as soon as possible! Much appreciated, thanks!EDIT: I've just found out and tested that vm.overcommit_memory=2 avoids the disk thrashing that leads to frozen OS (even though the OOM-killer doesn't get a change to trigger, which makes sense because it only triggered well after the disk-thrashing anyway, with vm.overcommit_memory=0) like: cc1plus: out of memory allocating 127440 bytes after a total of 897024 bytes
EDIT2 ok the above worked for me with the default vm.overcommit_ratio=50 but if I set it to 200 then the disk-thrashing is back!
EDIT3 ignore these 2 EDITs, they are not the way to fix this because processes that wouldn't have died before die now sooner, also vm.overcommit_ratio=0 will cause everything to die, or something. I'll try to find a better way, likely needing kernel recompile and I'm looking into the relevant GFP flags...
EDIT4: I found a way, through patching kernel, that works for me; see the patch inside this question.
answered Aug 19 at 10:23
Marcus Linsner
15914
add a comment |Â
up vote
1
down vote
I've found two explanations(of the same thing) as to why kswapd0 does constant disk reading happens well before OOM-killer kills the offending process:
- see the answer and comment of this askubuntu SE answer
- see the answer and David Schwartz's comments of this answer on unix SE
I'll quote here the comment from 1. which really opened my eyes as to why I was getting constant disk reading while everything was frozen:
For example, consider a case where you have zero swap and system is
nearly running out of RAM. The kernel will take memory from e.g.
Firefox (it can do this because Firefox is running executable code
that has been loaded from disk - the code can be loaded from disk
again if needed). If Firefox then needs to access that RAM again N
seconds later, the CPU generates "hard fault" which forces Linux to
free some RAM (e.g. take some RAM from another process), load the
missing data from disk and then allow Firefox to continue as usual.
This is pretty similar to normal swapping and kswapd0 does it. – Mikko
Rantalainen Feb 15 at 13:08
If anyone has a way as to how to disable this behavior(maybe recompile kernel with what options?), please let me know as soon as possible! Much appreciated, thanks!EDIT: I've just found out and tested that vm.overcommit_memory=2 avoids the disk thrashing that leads to frozen OS (even though the OOM-killer doesn't get a change to trigger, which makes sense because it only triggered well after the disk-thrashing anyway, with vm.overcommit_memory=0) like: cc1plus: out of memory allocating 127440 bytes after a total of 897024 bytes
EDIT2 ok the above worked for me with the default vm.overcommit_ratio=50 but if I set it to 200 then the disk-thrashing is back!
EDIT3 ignore these 2 EDITs, they are not the way to fix this because processes that wouldn't have died before die now sooner, also vm.overcommit_ratio=0 will cause everything to die, or something. I'll try to find a better way, likely needing kernel recompile and I'm looking into the relevant GFP flags...
EDIT4: I found a way, through patching kernel, that works for me; see the patch inside this question.
answered Aug 19 at 10:23
Marcus Linsner
15914
add a comment |Â
up vote
1
down vote
up vote
1
down vote
I've found two explanations(of the same thing) as to why kswapd0 does constant disk reading happens well before OOM-killer kills the offending process:
- see the answer and comment of this askubuntu SE answer
- see the answer and David Schwartz's comments of this answer on unix SE
I'll quote here the comment from 1. which really opened my eyes as to why I was getting constant disk reading while everything was frozen:
For example, consider a case where you have zero swap and system is
nearly running out of RAM. The kernel will take memory from e.g.
Firefox (it can do this because Firefox is running executable code
that has been loaded from disk - the code can be loaded from disk
again if needed). If Firefox then needs to access that RAM again N
seconds later, the CPU generates "hard fault" which forces Linux to
free some RAM (e.g. take some RAM from another process), load the
missing data from disk and then allow Firefox to continue as usual.
This is pretty similar to normal swapping and kswapd0 does it. – Mikko
Rantalainen Feb 15 at 13:08
If anyone has a way as to how to disable this behavior(maybe recompile kernel with what options?), please let me know as soon as possible! Much appreciated, thanks!EDIT: I've just found out and tested that vm.overcommit_memory=2 avoids the disk thrashing that leads to frozen OS (even though the OOM-killer doesn't get a change to trigger, which makes sense because it only triggered well after the disk-thrashing anyway, with vm.overcommit_memory=0) like: cc1plus: out of memory allocating 127440 bytes after a total of 897024 bytes
EDIT2 ok the above worked for me with the default vm.overcommit_ratio=50 but if I set it to 200 then the disk-thrashing is back!
EDIT3 ignore these 2 EDITs, they are not the way to fix this because processes that wouldn't have died before die now sooner, also vm.overcommit_ratio=0 will cause everything to die, or something. I'll try to find a better way, likely needing kernel recompile and I'm looking into the relevant GFP flags...
EDIT4: I found a way, through patching kernel, that works for me; see the patch inside this question.
answered Aug 19 at 10:23
Marcus Linsner
15914
I've found two explanations(of the same thing) as to why kswapd0 does constant disk reading happens well before OOM-killer kills the offending process:
- see the answer and comment of this askubuntu SE answer
- see the answer and David Schwartz's comments of this answer on unix SE
I'll quote here the comment from 1. which really opened my eyes as to why I was getting constant disk reading while everything was frozen:
For example, consider a case where you have zero swap and system is
nearly running out of RAM. The kernel will take memory from e.g.
Firefox (it can do this because Firefox is running executable code
that has been loaded from disk - the code can be loaded from disk
again if needed). If Firefox then needs to access that RAM again N
seconds later, the CPU generates "hard fault" which forces Linux to
free some RAM (e.g. take some RAM from another process), load the
missing data from disk and then allow Firefox to continue as usual.
This is pretty similar to normal swapping and kswapd0 does it. – Mikko
Rantalainen Feb 15 at 13:08
If anyone has a way as to how to disable this behavior(maybe recompile kernel with what options?), please let me know as soon as possible! Much appreciated, thanks!EDIT: I've just found out and tested that vm.overcommit_memory=2 avoids the disk thrashing that leads to frozen OS (even though the OOM-killer doesn't get a change to trigger, which makes sense because it only triggered well after the disk-thrashing anyway, with vm.overcommit_memory=0) like: cc1plus: out of memory allocating 127440 bytes after a total of 897024 bytes
EDIT2 ok the above worked for me with the default vm.overcommit_ratio=50 but if I set it to 200 then the disk-thrashing is back!
EDIT3 ignore these 2 EDITs, they are not the way to fix this because processes that wouldn't have died before die now sooner, also vm.overcommit_ratio=0 will cause everything to die, or something. I'll try to find a better way, likely needing kernel recompile and I'm looking into the relevant GFP flags...
EDIT4: I found a way, through patching kernel, that works for me; see the patch inside this question.
answered Aug 19 at 10:23
Marcus Linsner
15914
edited Aug 29 at 11:16
answered Aug 19 at 10:23
Marcus Linsner
15914
answered Aug 19 at 10:23
Marcus Linsner
15914
answered Aug 19 at 10:23
Marcus Linsner
15914
15914
add a comment |Â
add a comment |Â
Sign up or log in
StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
StackExchange.ready(
function ()
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2funix.stackexchange.com%2fquestions%2f373312%2foom-killer-doesnt-work-properly-leads-to-a-frozen-os%23new-answer', 'question_page');
);
Post as a guest
Sign up or log in
StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Sign up or log in
StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Post as a guest
Sign up or log in
StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
Sign up using Google
Sign up using Facebook
Sign up using Email and Password
1
I think this is what you should expect, if you're thrashing, but you're not really approaching 100% "used" i.e. there is too much memory usage which is file-backed, counted as "buff/cache". (Ugh, this phrasing assumes your tmpfs allocations are trivial, as these show up as "buff/cache", but cannot be paged out to a physical filesystem).
min_free_kbytesis not relevant, it's not a reserve for cached pages. AFAICT none of the vm sysctls allow reserving any memory specifically for cached pages, i.e. limiting MAP_ANONYMOUS allocations :(.– sourcejedi
Jun 28 '17 at 19:41
Thank you for having confirmed that setting another value to:
min_free_kbytesis not the right way to solve this issue. However, I disagree with your statement: "this is what you should expect". What I do expect is a system good enough to free the memory, by killing non vital processes, in high memory usage, and certainly not a frozen system, unusable for days.– parasite
Jun 29 '17 at 14:53
Ugh again. I was trying to make a technical statement, not a normative one. To be clear, I agree it's annoying there seems no way to reserve memory for cached pages, which you point out is the obvious way to try and mitigate this. I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process
– sourcejedi
Jun 29 '17 at 15:55
I'm not smart enough to say if there's a sensible way to account the necessary working set of cached pages to a process in a similar way to how anonymous pages are accounted to a process. I imagine the latter is easier... there's less ambiguity about whether the app is expected to use the page again later. Running out of memory sucks and Linux hasn't managed to mitigate that very much.
– sourcejedi
Jun 29 '17 at 16:01
1
I've been looking for a solution for this exact issue for years now without any success. I believe I first noticed the problem after replacing my HDD by an SSD, which also entailed me disabling swapping altogether, but I can't really guarantee that it never happened before these changes, so it might be unrelated. I'm on Archlinux btw.
– brunocodutra
Aug 8 '17 at 11:59