Andrew Bogott has uploaded a new change for review. ( 
https://gerrit.wikimedia.org/r/385939 )

Change subject: increase nfs mount timeouts
......................................................................

increase nfs mount timeouts

Puppet continues to flap occasionally on hosts that use nfs,
typically because a high IO load is causing the mounts
to timeout (which causes puppet to error out).

In many cases this IO slowness is a feature rather than a bug
(it's caused by per-instance throttling that we've put in on
purpose to prevent NFS overload).  So, let's just be patient
and try not to alert when things are slow.

Puppet has a full 30 minutes to complete a run so I wouldn't
mind increasing these even further, to 2 or even 5 minutes.

Change-Id: I2ccab28115fde1987e2781d99ed100e7ef43646a
---
M modules/labstore/manifests/nfs_mount.pp
M modules/toollabs/manifests/init.pp
2 files changed, 3 insertions(+), 3 deletions(-)


  git pull ssh://gerrit.wikimedia.org:29418/operations/puppet 
refs/changes/39/385939/1

diff --git a/modules/labstore/manifests/nfs_mount.pp 
b/modules/labstore/manifests/nfs_mount.pp
index 5ee1bad..6fc0af6 100644
--- a/modules/labstore/manifests/nfs_mount.pp
+++ b/modules/labstore/manifests/nfs_mount.pp
@@ -179,8 +179,8 @@
         # Puppet will normally get stuck and freeze raising load and 
effectively
         # failing to run
         exec { "create-${mount_path}":
-            command   => "/usr/bin/timeout -k 5s 10s /bin/mkdir ${mount_path}",
-            unless    => "/usr/bin/timeout -k 5s 30s /usr/bin/test -d 
${mount_path}",
+            command   => "/usr/bin/timeout -k 5s 20s /bin/mkdir ${mount_path}",
+            unless    => "/usr/bin/timeout -k 5s 60s /usr/bin/test -d 
${mount_path}",
             logoutput => true,
             require   => Mount[$mount_path],
         }
diff --git a/modules/toollabs/manifests/init.pp 
b/modules/toollabs/manifests/init.pp
index e859898..146df24 100644
--- a/modules/toollabs/manifests/init.pp
+++ b/modules/toollabs/manifests/init.pp
@@ -34,7 +34,7 @@
 
     exec {'ensure-grid-is-on-NFS':
         command => '/bin/false',
-        unless  => "/usr/bin/timeout -k 5s 30s /usr/bin/test -e 
${project_path}/herald",
+        unless  => "/usr/bin/timeout -k 5s 60s /usr/bin/test -e 
${project_path}/herald",
     }
 
     file { $sysdir:

-- 
To view, visit https://gerrit.wikimedia.org/r/385939
To unsubscribe, visit https://gerrit.wikimedia.org/r/settings

Gerrit-MessageType: newchange
Gerrit-Change-Id: I2ccab28115fde1987e2781d99ed100e7ef43646a
Gerrit-PatchSet: 1
Gerrit-Project: operations/puppet
Gerrit-Branch: production
Gerrit-Owner: Andrew Bogott <[email protected]>

_______________________________________________
MediaWiki-commits mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/mediawiki-commits

Reply via email to