MantisBT - Site
View Issue Details
0003529Site[All Projects] Bugpublic2018-10-01 04:442018-10-04 04:39
WubTheCaptain 
Blzut3 
immediatemajorsometimes
closedno change required 
0003529: Issue tracker write operations take unreasonably long time and are blocking I/O for that account
Since sometime today (possibly after 1000+ issues were closed), almost every write operation such as editing ticket details or adding a note takes around 3 minutes to finish, during which time the issue tracker is not usable except in incognito mode for read-only while not logged in.

Unbreak now.
No tags attached.
png 2018-10-01-045720_1267x868_scrot.png (471,956) 2018-10-01 05:01
/tracker/file_download.php?file_id=2418&type=bug
Issue History
2018-10-01 04:44WubTheCaptainNew Issue
2018-10-01 04:56WubTheCaptainNote Added: 0019853
2018-10-01 05:01WubTheCaptainFile Added: 2018-10-01-045720_1267x868_scrot.png
2018-10-01 05:03WubTheCaptainNote Added: 0019855
2018-10-01 05:03Blzut3Note Added: 0019856
2018-10-01 05:05WubTheCaptainNote Added: 0019857
2018-10-02 05:35WubTheCaptainNote Added: 0019879
2018-10-03 02:19Blzut3Assigned To => Blzut3
2018-10-03 02:19Blzut3Statusnew => assigned
2018-10-03 23:09Blzut3Note Added: 0019896
2018-10-04 00:01Blzut3Note Added: 0019897
2018-10-04 03:27Blzut3Note Added: 0019900
2018-10-04 03:27Blzut3Statusassigned => closed
2018-10-04 03:27Blzut3Resolutionopen => no change required
2018-10-04 04:33WubTheCaptainNote Added: 0019901
2018-10-04 04:33WubTheCaptainStatusclosed => feedback
2018-10-04 04:33WubTheCaptainResolutionno change required => reopened
2018-10-04 04:37Blzut3Note Added: 0019902
2018-10-04 04:37Blzut3Statusfeedback => closed
2018-10-04 04:37Blzut3Resolutionreopened => no change required
2018-10-04 04:39Blzut3Note Edited: 0019902bug_revision_view_page.php?bugnote_id=19902#r12095

Notes
(0019853)
WubTheCaptain   
2018-10-01 04:56   
I've been enduring this for few hours and thought to wait for it to go away, but a ticket is better than no ticket.

This issue is also visible in issue histories, such as 0003483's:

2018-10-01 06:56 	WubTheCaptain		Relationship deleted 	related to 0003479
2018-10-01 06:59 	WubTheCaptain		Relationship deleted 	related to 0003482
2018-10-01 07:01 	WubTheCaptain		Relationship deleted 	related to 0003505
(0019855)
WubTheCaptain   
2018-10-01 05:03   
Also, the "last update" date is severely wrong but probably unrelated to this issue (as in attachment, which demonstrates this and the issue tracker being very slow to respond). Although, in browser incognito mode the notes and updates go live much faster, it still makes that account unusable for a good while.
(0019856)
Blzut3   
2018-10-01 05:03   
It should clear up in 24 hours. That said if you are going to make a ticket like this you should probably refrain from "unbreak now." :P
(0019857)
WubTheCaptain   
2018-10-01 05:05   
Quote from Blzut3
you should probably refrain from "unbreak now." :P


It's a priority level at Wikimedia Foundation (in MediaWiki software development, for example). :P

'https://www.mediawiki.org/w/index.php?title=Phabricator/Project_management&oldid=2865675#Priority_levels [^]'
(0019879)
WubTheCaptain   
2018-10-02 05:35   
Progress report? The outbound email server (my guess) is still clogged.
(0019896)
Blzut3   
2018-10-03 23:09   
Based on the emails I'm getting today, I think surprisingly it's working through the backlog. I thought it would just drop them.
(0019897)
Blzut3   
2018-10-04 00:01   
OK learned about Mantis's email que and cleared out the remaining 222 emails about tickets being closed.
(0019900)
Blzut3   
2018-10-04 03:27   
Seems to be clear now, if not it will definitely clear up very soon.
(0019901)
WubTheCaptain   
2018-10-04 04:33   
I don't think this issue is seemingly resolved from what I can read from the notes. "No change required" is not the right solution here either, you've done something to have temporarily "fixed" it. Someday someone may inevitably run into this issue again, and the blame may be on MantisBT's code or the existing email infrastructure.

I've said "unbreak now" in the OP for a reason.

Quote from Blzut3
I think surprisingly it's working through the backlog. I thought it would just drop them.


MTAs never drop email, they must accept it for delivery or generate a bounce as designed. "It MUST NOT lose the message for frivolous reasons, such as because the host later crashes or because of a predictable resource shortage." This is hardly surprising to me. (RFC 5321, section 6.1)

Besides: Have you tried OpenSMTPD for a local MTA to relay messages (like an MSA)? Please do. What steps will the site administration take to prevent this issue from happening again?
(0019902)
Blzut3   
2018-10-04 04:37   
(edited on: 2018-10-04 04:39)
I put no change required since I did nothing but sit back and relax. (Well except flushing the queue, but lets be real here.)

The software is doing what it's supposed to do, we're just using free services and they have limits. There's nothing to do here.

Un-pedant now. :P

Edit: and at the typical volume of traffic that we have I have better things to do with my time then setup an SMTP server.