Project

General

Profile

Bug #1018

vnrlu_proc: vnode recycler stopped working!

Added by mneumann about 10 years ago. Updated about 10 years ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Category:
-
Target version:
-
Start date:
Due date:
% Done:

0%

Estimated time:

Description

I'm getting the same problem as described in [1] with the most recent
head revision. I just tried to "cvs checkout" and then the message
appears repetitive and no further progress of cvs checkout seems to
happen.

Regards,

Michael

[1]: http://leaf.dragonflybsd.org/mailarchive/bugs/2004-12/msg00020.html

History

#1 Updated by dillon about 10 years ago

:I'm getting the same problem as described in [1] with the most recent
:head revision. I just tried to "cvs checkout" and then the message
:appears repetitive and no further progress of cvs checkout seems to
:happen.
:
:Regards,
:
: Michael
:
:[1]: http://leaf.dragonflybsd.org/mailarchive/bugs/2004-12/msg00020.html

That's a really old bug that was fixed a long time ago, or should have
been.

I did recently (a few weeks ago) make some changes to try to prevent
the buffer cache deadlock from occuring, perhaps I muffed it up. I
haven't touched that part of the codebase it in the last week, though.

Can you generate a kernel core (& binary) of the affected machine and
upload it into ~/crash on your leaf account? I'll take a look at it.

-Matt
Matthew Dillon
<>

#2 Updated by pavalos about 10 years ago

I got this today when trying to cvs update my pkgsrc tree. I'm
uploading a crashdump right now (*.14).

--Peter

#3 Updated by dillon about 10 years ago

Were both of you guys doing major filesystem operations on the root
mount?

From looking at Peter's core the root mount point doesn't have a syncer
vnode, and the new vnode recycling code won't run on it because of that.
I am fixing it now.

What is interesting is that the vnode recycling code is not in error
here, but instead has exposed a long standing issue in the code where
the syncer was not running on the root filesystem every 30 seconds like
it was supposed to.

-Matt

#4 Updated by pavalos about 10 years ago

Yes, I think both of us mentioned we were doing cvs checkouts/updates.
Additionally my mail server was processing a good amount of email.

--Peter

#5 Updated by dillon about 10 years ago

:Yes, I think both of us mentioned we were doing cvs checkouts/updates.
:Additionally my mail server was processing a good amount of email.
:
:--Peter

Try it w/ that fix I committed.

-Matt

#6 Updated by mneumann about 10 years ago

Your commit fixes the problem. I close the bug report.

Also available in: Atom PDF