PostgreSQL and the OOM killer: Why we use strict memory overcommit

Join the discussion

• Hacker News
• There's so much great stuff here.

  First, Linux's default memory management strategy is bonkers. OOM killing rarely actually works in my experience, at least on desktop. It takes ages to kick in and usually the system just freezes and you have to hard reboot. I've experienced this on every Linux system I've used, even my current one with 128GB of RAM and 64GB of swap, so don't say "it works for me". Windows and Mac do not have this issue at all, so clearly it's possible to do it better.

  Has anyone tried using strict overcommit on desktop Linux?

  Second, this bug is a great counterpoint to those annoying people who naysay Rust with "but not all bugs are memory safety bugs, what about logic bugs? huh?". Rust code would not have had this bug.

  by IshKebab
• What exactly does Rust solve here? Virtual memory is a hardware/OS feature.
  by grg0
• Yes, many have tried to use strict overcommit on the desktop. It is a good footgun. https://unix.stackexchange.com/a/797888/1027
  by man8alexd
• I have disabled overcommit both on Windows and on Linux. I hate having random programs being killed.

  Unfortunately, many programs commit 2x memory than they actually use. Often I see ~32GB committed and ~16GB resident.

  by szmarczak
• how exactly did you disabled it on Windows?

  I dont think it has an option for that.

  by nok22kon
• disabling overcommit is trading OOM for random program crashes due to the inability to handle ENOMEM. It also wastes a lot of system memory.

  https://unix.stackexchange.com/questions/797835/disabling-ov...

  by man8alexd
• Does this result in programs more frequently erroring/crashing because they can't allocate? I don't know how well many of the programs I frequently use on my desktop (Firefox, GNOME desktop, JVM + IntelliJ, Slack, etc.) handle allocation failures. I'm not sure they would do much better than crash, but I know the default OOM killer settings work well for me. About once a year a real runaway process (usually a throwaway program I'm working on) gets OOM-killed, and that's fine with me.
  by sterwill
• Mode 0 (Heuristic) is described incorrectly. All this complex heuristic was removed almost a decade ago. Currently, the kernel refuses a single allocation that exceeds the physical memory. That is all.

  The article ignores the proper modern solution to prevent OOM killing of critical processes - OOM Score Adjust.

  Tuning CommitLimit manually is an archaic, imprecise, and error-prone way to handle memory limits, only suitable for single-process workloads that can handle ENOMEM properly. It completely ignores dynamic file page cache memory allocation. You still can get OOM if you get unusually high file activity. On the other hand, under low file activity, it wastes memory on the same page cache, because it can't be reclaimed without memory pressure, and memory pressure can't be created because workload hits ENOMEM earlier. Don't use strict overcommit.

  by man8alexd
• The key thing is Postgres does handle enomem well and does a nice rollback rather than crashing the server and entering crash recovery. It’s one of the few programs that does. Exceptions for the exception.

  Even a revised heuristic that only spots large, individual allocations is not going to do the job.

  Oom score adjust also doesn’t do the job: because the only interesting workload is Postgres, if a backend does a page fault that needs memory, who dies? Another sibling Postgres, almost certainly. Then postmaster does crash recovery, which most would rather avoid. High performance databases with distant checkpoints can take a while to come back up.

  by fdr
• Nothing worse than memory management on Hyperscaler VMs which do not use Swap :|

  Took k8s ages to get Swap support.

  We lost something when we accepted that Hyperscalers just tell you to use more moemory. It was shitty 5 years ago and today especially after the ram price increases

  by chiply314
• My guess would be: it's because memory management before MGLRU was really not good and required different userspace solutions and tinkering. You either get killed with OOM (no swap) or got into thrashing (swap).

  And now, with PSI + MGLRU, situation is much better, but there are still missing features/subsystems which would be nice to have. For example there's no simple way to lock memory mlockall-style to ensure that rarely used daemon would not face long no-cache-latency upon accessing the first time after long idle time.

  by ValdikSS
• For once, Microsoft's decision to just not do overcommit in Windows seems sensible
  by wongarsu
• Windows doesn't need to fork, and you can't fork a large process without overcommit.
  by man8alexd
• Microsoft just followed VAX/VMS that does not overcommit. And there is a noise on Linux mail lists to implement process builder pattern which VAX had like 50 years ago…
  by fpoling
• I read this article about 3 weeks ago when this bit me. Really great write-up, some tricky details.
  by adamors
• The proper way to handle OOM is to do what mature databases do: implement your own memory accounting, use only your own allocators integrated with the accounting system, and ensure that every allocation path can recover from OOM. Easier said than done.
  by senderista
• MariaDB recently implemented memory PSI monitoring but failed with that in a curious way and disabled it afterwards by default. The failure is that under memory pressure, they flushed the entire InnoDB buffer pool.
  by man8alexd
• I have a couple of points, not really sure if they should be in one post or not, but whatever.

  Firstly, if you take what's written at face value, it seems there's a serious logic error in the OOM killer. If it genuinely counts shared memory against a process, then its logic is wrong because killing that process wouldn't release that much memory, it'd need to kill all the processes sharing that memory to release it. So, maybe it should ignore shared memory in its calculations, or weight them by number of processes sharing it, or whatever.

  The other issue is kind of true, but shows a stubbornness from the developers to change how they approach the problem. It's true that if any task could be partially through updating shared memory when it is killed, then all bets are off as to the state of that memory. However, if that's the case, then there should already be some kind of locking mechanisms in place to prevent multiple processes updating the same pages anyway.

  Probably the current solution is: something gets locked when modified; every other process would need to spinlock until it's released; locking process is killed; everything else is stuck; another PG thread notices the child has died and kills everything else; on restart the DB has to be recovered.

  A different solution could be: give every process its own private part of the shared memory for when it starts a transaction; have an indirection table from page number to shared memory page; for every page that needs to be modified, a new page is allocated from the freed pages list; that allocation is recorded in the private part of the shared memory along with the page number it's replacing; the old page is left untouched and copied to the new page along with any changes; the change list is terminated; then as the last step we update the indirection table for every page that was modified. If the process was killed at any point, we can either roll back the entirety of the transaction (freeing every allocation it made for replacement pages), or if the list was marked as terminated, finish off updating the indirection table with the changes required and marking the original pages as unused. At that point, the lock can be released knowing that shared memory is still entirely consistent.

  Some of that is probably happening anyway if postgres supports reading from tables concurrently with an active write transaction on the same table, in which case the logic on how to mark those now freed pages as still in use is required. In that case, each process can also maintain a list of pages it has marked as still being used in case a reading process is killed off.

  by ralferoo
• About shared memory included in the memory size calculations https://lkml.iu.edu/1902.2/05674.html
  by man8alexd
• I'd be interested to see a Linux distribution whose entire shtick is to run well-behaved under a kernel with overcommit disabled. But it would be a huge undertaking. Besides the obvious issue with fork(), there are a lot of programs and libraries out there that implicitly rely on overcommit due to not checking malloc() for failure.
  by 10000truths
• Sigh. Malloc failure should have had to be trapped with a signal or something, not just a return status. I know, I know, threads and nesting handlers make that hard and historical precedent makes it impossible to retrofit, but I can still dream.
  by zbentley
• There is some kind of illusion or myth that strict overcommit solves memory management issues.
  by man8alexd
• I think this is also a good lesson on why it's best to isolate mission-critical services like databases on their own compute nodes.
  by otterley
• The problem with disabling the memory overcommit is that then the RAM is wasted. That can be worked around with setting up swap but then the disk space is wasted.
  by mono442
• Why are programs even allocating memory that they don't use?
  by frollogaston
• People like to reinvent things that they are not aware of. Original BSDs used to use strict swap reservation - every anonymous memory page had to have an associated swap page. You had to have the swap 2x of RAM to allow large processes to fork - otherwise you would get an "out of swap" error. FreeBSD implemented overcommit around 2000, I think version 4.x or 5.x.
  by man8alexd
• This has bitten me multiple times. The problem I have is that at work we deploy the application (written in Go) and PostgreSQL on the same machine. The backend app allocates a lot of virtual memory, and initially we had overcommit to 0 (heuristic). This caused crashes on big queries in PostgreSQL and we set it to 2. The whole system became a bit unstable because the backend would still allocate a lot of virtual memory and at some point we ran into errors when allocating.

  For now, we have overcommit_ratio set to a value that is stable from experience, but there really seems to be no silver lining. Go is very happy to allocate a lot of virtual memory, but so are most managed languages. The best solution would probably be to host the backend and the database on separate servers.

  by leononame
• Yes, it would. Basically every serious database tries to allocate everything and more - back in the day we'd just allocate VMs on the machine even with the overhead because knowing it cannot leave its constraints and would work within them was worth the cost.
  by hilariously
• You can also use cgroup to set allocation policy for a particular program.
  by fpoling
• I'm not sure if you are aware but there are relatively recent environment variables you can set to help contain Go memory to a fixed size.

  GOMEMLIMIT works very well if you set it to around 90% of available memory as a rough heuristic. You should definitely profile your application to fine tune this number (e.g. if you link with C libraries that hold large memory pools then Go doesn't account for that) but also to identify sources of spikey/leaky allocations. For example, encoding/json is notorious for it's inner sync.Pool hanging on to outsized buffers. There's usually a lot of low hanging fruit.

  In my experience Go can be extremely stable in terms of memory footprint at both small (~O(1MiB)) and large (~O(256GiB)) scales, and it takes only a small amount of effort.

  As far as GC languages go, it is by far the easiest to work with.

  by xyzzy_plugh