#1750: Fail more gracefully on -l >= 4GB
---------------------------------+----------------------
Reporter: geoff | Type: defect
Status: new | Priority: low
Milestone: Varnish 4.0 release | Component: varnishd
Version: 4.0.3 | Severity: minor
Keywords: |
---------------------------------+----------------------
Varnish cannot run with a VSM file size of 4GB or more -- as discussed
with Martin on IRC, the VSM API cannot accommodate sizes that large (for
example due to unsigned ints as offsets).
But the varnishd command succeeds with -l 4g -- the management process
starts, and then the child process fails an assert. So the invocation
appears to be normal, although Varnish fails almost immediately.
On my system, there was a message in /var/log/messages indicating failure
of the child due to SIGABRT, but there was no core dump (although I always
get them otherwise). And since the management process also stopped, there
was no panic message, so there was nothing useful to diagnose the problem.
I suggest checking -l against the maximum on invocation, and failing right
away to stderr.
--
Ticket URL: <https://www.varnish-cache.org/trac/ticket/1750>
Varnish <https://varnish-cache.org/>
The Varnish HTTP Accelerator
_______________________________________________
varnish-bugs mailing list
varnish-bugs@varnish-cache.org
https://www.varnish-cache.org/lists/mailman/listinfo/varnish-bugs
---------------------------------+----------------------
Reporter: geoff | Type: defect
Status: new | Priority: low
Milestone: Varnish 4.0 release | Component: varnishd
Version: 4.0.3 | Severity: minor
Keywords: |
---------------------------------+----------------------
Varnish cannot run with a VSM file size of 4GB or more -- as discussed
with Martin on IRC, the VSM API cannot accommodate sizes that large (for
example due to unsigned ints as offsets).
But the varnishd command succeeds with -l 4g -- the management process
starts, and then the child process fails an assert. So the invocation
appears to be normal, although Varnish fails almost immediately.
On my system, there was a message in /var/log/messages indicating failure
of the child due to SIGABRT, but there was no core dump (although I always
get them otherwise). And since the management process also stopped, there
was no panic message, so there was nothing useful to diagnose the problem.
I suggest checking -l against the maximum on invocation, and failing right
away to stderr.
--
Ticket URL: <https://www.varnish-cache.org/trac/ticket/1750>
Varnish <https://varnish-cache.org/>
The Varnish HTTP Accelerator
_______________________________________________
varnish-bugs mailing list
varnish-bugs@varnish-cache.org
https://www.varnish-cache.org/lists/mailman/listinfo/varnish-bugs