Hi Anthony. We were using RAID + EXT4, with journalling enabled.
Thanks for providing the link to the Mark Price blog .. those articles were fantastic in helping me understand the issue. We moved several of the Kafka servers to XFS + JBOD and no longer get the large pauses. Thanks for your help, Ross On Friday, 7 December 2018 20:53:43 UTC+11, Anthony Maire wrote: > > What file system are you using ? > EXT4 journalling could really be a huge perf killer in such usecase and > cause this TTSP issue if Kafka uses mmap ( cf > http://epickrram.blogspot.com/2015/12/journalling-revisited.html : After > much detailed investigation, we tracked the problem down to interaction > between the writer code and the ext4 journalling daemon > <http://lxr.free-electrons.com/source/fs/ext4/ext4_jbd2.c?v=3.18> (jbd2) ) > If you are using EXT4, you might want to do some tests with XFS or > disabling journaling, it might help. > > >> -- You received this message because you are subscribed to the Google Groups "mechanical-sympathy" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
