Color Cloud
2015-09-21 22:23:27 UTC
Hello,
We are synchronizing an AWS instance directory with a bucket using s3cmd
and lsyncd.
We have had some cases where s3cmd eats all CPU from the instance (during
normal sync process). We upgraded to version 1.5.1.2, but still happens
sometimes (previous version was
s3cmd-1.5.0-0.7.rc1.el6.noarch)
One i captured the process that got stuck, and the file that was being
synchronized was smaller than others that were synchronized without
problem.
We receive an alarm of cpu-credit, and check the instance with top, s3cmd
is using all of the processor, and sending a SIGKILL to the stuck process
"fixes" s3cmd process and the next files continue synchronizing correctly.
How we can debug the cause of this problem?
CentOS 6.6
s3cmd 1.5.1.2
lsyncd 2.1.5
Directories being watched by lsyncd 31018
Thanks a lot!
We are synchronizing an AWS instance directory with a bucket using s3cmd
and lsyncd.
We have had some cases where s3cmd eats all CPU from the instance (during
normal sync process). We upgraded to version 1.5.1.2, but still happens
sometimes (previous version was
s3cmd-1.5.0-0.7.rc1.el6.noarch)
One i captured the process that got stuck, and the file that was being
synchronized was smaller than others that were synchronized without
problem.
We receive an alarm of cpu-credit, and check the instance with top, s3cmd
is using all of the processor, and sending a SIGKILL to the stuck process
"fixes" s3cmd process and the next files continue synchronizing correctly.
How we can debug the cause of this problem?
CentOS 6.6
s3cmd 1.5.1.2
lsyncd 2.1.5
Directories being watched by lsyncd 31018
Thanks a lot!