Production postmortemThe random high CPU

time to read 2 min | 253 words

A customer complained that every now and then RavenDB is hitting 100% CPU and stays there. They were kind enough to provide a minidump, and I started the investigation.

I loaded the minidump to WinDB and started debugging. The first thing you do with high CPU is rung the “!runaway” command, which sorts the threads by how busy they are:

image

I switched to the first thread (39) and asked for its stack, I highlighted the interesting parts:

image

This is enough to have a strong suspicion on what is going on. I checked some of the other high CPU threads and my suspicion was confirmed, but even from this single stack trace it is enough.

Pretty much whenever you see a thread doing high CPU within the Dictionary class it means that you are accessing it in a concurrent manner. This is unsafe, and may lead to strange effects. One of them being an infinite loop.

In this case, several threads were caught in this infinite loop. The stack trace also told us where in RavenDB we are doing this, and from there we could confirm that indeed, there is a rare set of circumstances that can cause a timer to fire fast enough that the previous timer didn’t have a chance to complete, and both of these timers will modify the same dictionary, causing the issue.

More posts in "Production postmortem" series:

  1. (23 Mar 2020) high CPU when there is little work to be done
  2. (21 Feb 2020) The self signed certificate that couldn’t
  3. (31 Jan 2020) The slow slowdown of large systems
  4. (07 Jun 2019) Printer out of paper and the RavenDB hang
  5. (18 Feb 2019) This data corruption bug requires 3 simultaneous race conditions
  6. (25 Dec 2018) Handled errors and the curse of recursive error handling
  7. (23 Nov 2018) The ARM is killing me
  8. (22 Feb 2018) The unavailable Linux server
  9. (06 Dec 2017) data corruption, a view from INSIDE the sausage
  10. (01 Dec 2017) The random high CPU
  11. (07 Aug 2017) 30% boost with a single line change
  12. (04 Aug 2017) The case of 99.99% percentile
  13. (02 Aug 2017) The lightly loaded trashing server
  14. (23 Aug 2016) The insidious cost of managed memory
  15. (05 Feb 2016) A null reference in our abstraction
  16. (27 Jan 2016) The Razor Suicide
  17. (13 Nov 2015) The case of the “it is slow on that machine (only)”
  18. (21 Oct 2015) The case of the slow index rebuild
  19. (22 Sep 2015) The case of the Unicode Poo
  20. (03 Sep 2015) The industry at large
  21. (01 Sep 2015) The case of the lying configuration file
  22. (31 Aug 2015) The case of the memory eater and high load
  23. (14 Aug 2015) The case of the man in the middle
  24. (05 Aug 2015) Reading the errors
  25. (29 Jul 2015) The evil licensing code
  26. (23 Jul 2015) The case of the native memory leak
  27. (16 Jul 2015) The case of the intransigent new database
  28. (13 Jul 2015) The case of the hung over server
  29. (09 Jul 2015) The case of the infected cluster