[
https://issues.apache.org/jira/browse/CASSANDRA-12699?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15525576#comment-15525576
]
Heiko Sommer commented on CASSANDRA-12699:
------------------------------------------
I think that C* could have page tables much smaller than the 3-4 GB that I see,
no matter how large the data files are. For compactions and anticompactions, I
believe that C* reads through large data files, merging or splitting their
content in an ordered fashion, but never jumping around in any of these files.
Shouldn't it be possible to "mmap" only a small "running working section" of
these files, and to quickly "munmap" the pages that have been processed
already? Or to "mmap" the whole file at once, trusting that Linux will only
populate the page table upon page access (which explains the slow ramp-up of
PTE memory in my plot), but still to "munmap" pages as soon as possible?
This immediate unmapping might be difficult to implement, especially if Java
libraries do not support it, but at least
http://lanyrd.com/2013/apachecon/scctbq/ talks about C* already doing unmapping
tricks. Could it be extended to the file access done for (anti-)compactions?
I agree that buying 4 GB more RAM to work around the PTE memory issue could be
an easy option for our cluster, but I still suspect that there could be "freak
wave" cases where much more than 4 GB PTE are needed, so that the issue is of
general interest to get it fixed.
I will also look into issues with VmData and anonymously mapped memory that
Ariel pointed out, to see how our cluster is doing with these.
> Excessive use of "hidden" Linux page table memory
> -------------------------------------------------
>
> Key: CASSANDRA-12699
> URL: https://issues.apache.org/jira/browse/CASSANDRA-12699
> Project: Cassandra
> Issue Type: Bug
> Components: Core
> Environment: Cassandra 2.2.7 on Red Hat 6.7, with Java 1.8.0_73.
> Probably others.
> Reporter: Heiko Sommer
> Attachments: PageTableMemoryExample.png, cassandra-env.sh,
> cassandra.yaml, cassandraMemoryLog.sh, cassandraMemoryLog.sh
>
>
> free
> The cassandra JVM process uses many gigabytes of page table memory during
> certain activities, which can lead to oom-killer action with
> "java.lang.OutOfMemoryError: null" logs.
> Page table memory is not reported by Linux tools such as "top" or "ps" and
> therefore might be responsible also for other spurious Cassandra issues with
> "memory eating" or crashes, e.g. CASSANDRA-8723.
> The problem happens especially (or only?) during large compactions and
> anticompactions.
> Eventually all memory gets released, which means there is no real leak. Still
> I suspect that the memory mappings that fill the page table could be released
> much sooner, to keep the page table size at a small fraction of the total
> Cassandra process memory.
> How to reproduce: Record the memory use on a Cassandra node, including page
> table memory, for example using the attached script cassandraMemoryLog.sh.
> Even when there is no crash, the ramping up and sudden release of page table
> memory is visible.
> A stacked area plot for the memory on one of our crashed nodes is attached
> (PageTableMemoryExample.png). The page table memory used by Cassandra is
> shown in red ("VmPTE").
> (In the plot we also see that the sum of measured memory portions sometimes
> exceeds the total memory. This is probably an issue of how RSS memory is
> measured, perhaps including some buffers/cache memory that also counts toward
> available memory. It does not invalidate the finding that page table memory
> is growing to enormous sizes.)
> Shortly before the crash, /proc/$PID/status reported
> VmPeak: 6989760944 kB
> VmSize: 5742400572 kB
> VmLck: 4735036 kB
> VmHWM: 8589972 kB
> VmRSS: 7022036 kB
> VmData: 10019732 kB
> VmStk: 92 kB
> VmExe: 4 kB
> VmLib: 17584 kB
> VmPTE: 3965856 kB
> VmSwap: 0 kB
> The files cassandra.yaml and cassandra-env.sh used on the node where the data
> was taken are attached.
> Please let me know if I should provide any other data or descriptions to help
> with this ticket.
> Known workarounds: Use more RAM, or limit the amount of Java heap memory. In
> the above crash, MAX_HEAP_SIZE was not set, so that the default heap size for
> 12 GB RAM was used (-Xms2976M, -Xmx2976M).
> We have not tried yet if variations of heap vs. offheap config choices make a
> difference.
> Perhaps there are other workarounds using -XX+UseLargePages or related Linux
> settings to reduce the size of the process page table?
> I believe that we see these crashes more often than other projects because we
> have a test system with not much RAM but with a lot of data (compressed ~3 TB
> per node), while the CPUs are slow so that anti-/compactions overlap a lot.
> Ideally Cassandra (native) code should be changed to release memory in
> smaller chunks, so that page table size cannot cause an otherwise stable
> system to crash.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
