[Gluster-users] Long response time on idle volume

[Jan Wrona]

Hello,

I'm using one volume exclusively for configuration files, therefore 
there is no I/O within this volume when the cluster is up and running. 
It is a replicate volume with three bricks, one on each node. Pacemaker 
is used to manage services including Gluster, using the 
ocf::heartbeat:Filesystem resource agent to manage mounts on the client 
side. This RA offers three monitor levels (0, 10, 20), I'm having 
problem with the monitor of level 20, which basically just writes a 
small status file to the volume and reads it afterward. Monitoring 
interval is 60 seconds and timeout is 20 seconds. Most of the time 
everything work just fine the monitor operation takes about a second and 
terminates successfully, but sometimes it times out (which makes 
Pacemaker think that Gluster mount failed).

My question is why it suddenly takes more than 20 seconds to write few 
bytes to the volume? Is there anything I can do with this (other than 
prolonging the timeout)?

Many thanks,
JW

-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 3706 bytes
Desc: S/MIME Cryptographic Signature
URL: <http://lists.gluster.org/pipermail/gluster-users/attachments/20170320/9ee36804/attachment.p7s>