[Archivesspace_Users_Group] Connection/login dropping

[RENTON Scott]

Hi all

Just to confirm, Blake and Mark were able to resolve this issue, as we're not the first to hit it, so thanks. It looks as though it was down to a failed merging of top_containers, which left some orphaned sub_containers (with null top_container references in the top_containers_link_rlshp table). We knew we had this issue but thought it was just affecting those records- however, it does seem to have had the knock on effect of causing login to fail.

If anyone else hits it, the fix is here: https://gist.github.com/mark-cooper/bf8b9d5e86255b92b9576f683206ba79

Cheers
Scott


==========

Scott Renton

Digital Library Development & Systems

Floor F East

Argyle House

515219

________________________________
From: RENTON Scott
Sent: 06 September 2021 14:29
To: archivesspace_users_group at lyralists.lyrasis.org <archivesspace_users_group at lyralists.lyrasis.org>
Subject: Connection/login dropping

Hi all

Our uptime has been a bit flaky the last few weeks- what we're finding is that, after a certain amount of service, users are unable to login, and if that's not noticed or acted on, the site generally stops responding some time later. A restart of the application resolves it.

We're running 2.8.1 against MariaDB 5.5-65, and Java 8. I have tried different versions of the mysqlconnector, but this does not seem to make a difference (first noticed on 5.1.39). The MySQL has been tuned to an extent. The VM has 16GB of memory, approximately 10GB of which is apportioned to AS. The MySQL process was restarted last week, and clearing that has still allowed the issue.

It's pretty hard to spot any particular action which causes this. The logging we see once it's happened looks like this:

E, [2021-08-18T06:21:01.548843 #90658] ERROR -- : Thread-28310: Unhandled exception!
E, [2021-08-18T06:21:01.552325 #90658] ERROR -- :

apps.archivesspace.gems.gems.sequel_minus_5_dot_9_dot_0.lib.sequel.model.base.RUBY$method$parent_id$0(/apps/archivesspace/gems/gems/sequel-5.9.0/lib/sequel/model/base.rb:783)
apps.archivesspace.gems.gems.sequel_minus_5_dot_9_dot_0.lib.sequel.model.base.RUBY$method$parent_id$0$__VARARGS__(/apps/archivesspace/gems/gems/sequel-5.9.0/lib/sequel/model/base.rb)
org.jruby.internal.runtime.methods.CompiledIRMethod.call(CompiledIRMethod.java:82)
org.jruby.internal.runtime.methods.MixedModeIRMethod.call(MixedModeIRMethod.java:70)

and will continue to spout "minus_​5_dot_9_dot_0" type errors until we reboot.

I have taken some statistics around the numbers of threads running pre and post bounce. I don't see anything obvious there, and I have left the config.rb value

AppConfig[:db_max_connections] = proc { 20 + (AppConfig[:indexer_thread_count] * 2) }


as it is.

I'm also aware that we have an issue some others have of a proliferation of FATAL errors since merging top/subcontainers (items cannot be found), but I do not know if it is related to this.

However, if anyone recognises this behaviour, and can advise, I'd be grateful to hear about it.

Cheers
Scott




==========

Scott Renton

Digital Library Development & Systems

Floor F East

Argyle House

515219

The University of Edinburgh is a charitable body, registered in Scotland, with registration number SC005336. Is e buidheann carthannais a th’ ann an Oilthigh Dhùn Èideann, clàraichte an Alba, àireamh clàraidh SC005336.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lyralists.lyrasis.org/pipermail/archivesspace_users_group/attachments/20211013/507d3816/attachment.html>