warning: log line attempted over max size - leadership related

Bug #1457645 reported by David Britton
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
juju-core
Fix Released
High
Ian Booth
1.23
Fix Released
Critical
Ian Booth
1.24
Fix Released
Critical
Ian Booth

Bug Description

This environment is nothing too special. Juju 1.23.0, been running for over a month. Now I'm getting lots of errors in my syslog, meanwhile jujud and mongo are both spinning on cpu taking up in the 60% range all the time, while using up in the 1G range of memory.

Restarting the services is not enough to clear things up.

I attached a full machine-0.log (it's over 9G uncompressed) and a mongo database dump.

Getting messages repeating every minute or so like this:

May 21 18:44:44 juju-lcy01-machine-0 mongod.37017[7124]: Thu May 21 18:44:44.077 [conn50] warning: log line attempted (677k) over max size(10k), printing beginning and end ... update juju.txns.stash query: { _id: { c: "lease", id: "trusty-server-leadership" }, n: "d5b7100b" } update: { $set: { txn-queue: [ "55352def91a7b10f460007b4_127104b4", "55352e1091a7b10f46000802_01edfd48", "55352e2a91a7b10f4600085a_dbb7318e", "55352e4891a7b10f460008ce_01a667f7", "55352e6591a7b10f460009cd_3991da34",
...
"555e26bc91a7b11ba5029c5f_113c738e", "555e271f91a7b11ba5029c6e_9c81bef2", "555e279591a7b11ba5029c7d_06584eb9" ], txn-revno: -37441 }, $unset: { n: 1 } } nscanned:1 nmoved:1 nupdated:1 keyUpdates:0 locks(micros) w:246896 126ms

Revision history for this message
David Britton (dpb) wrote :
Revision history for this message
David Britton (dpb) wrote :
David Britton (dpb)
tags: added: cloud-installer
Curtis Hovey (sinzui)
description: updated
tags: added: mongodb
Curtis Hovey (sinzui)
Changed in juju-core:
milestone: none → 1.25.0
status: New → Triaged
importance: Undecided → High
Revision history for this message
Menno Finlay-Smits (menno.smits) wrote :

I've had a look at the DB and the lease collection looks like it has problems. All the lease documents have huge numbers of txn-queue entries, which all refer to completed transactions. Typically mgo/txn will leave only the last completed transaction to touch a document in the txn-queue field so this is quite unusual.

Here's the txn-queue sizes of each of the lease docs in the dump:

landscape-client-precise-leadership: 20746
landscape-client-trusty-leadership: 21381
landscape-client-utopic-leadership: 20685
landscape-client-vivid-leadership: 5942
precise-client-leadership: 18785
precise-server-leadership: 18777
swap-leadership: 18690
trusty-client-leadership: 18769
trusty-server-leadership: 18771
ubuntu-leadership: 18782
utopic-client-leadership: 18747
utopic-server-leadership: 18751
vivid-client-leadership: 18714
vivid-server-leadership: 5918

I'm not sure if this is related to bug 1454891 but it certainly could be.

Ian Booth (wallyworld)
Changed in juju-core:
assignee: nobody → Ian Booth (wallyworld)
importance: High → Critical
status: Triaged → In Progress
Ian Booth (wallyworld)
Changed in juju-core:
status: In Progress → Fix Committed
Curtis Hovey (sinzui)
Changed in juju-core:
importance: Critical → High
status: Fix Committed → Fix Released
David Britton (dpb)
tags: added: kanban-cross-team
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.