Pulling other objects sometimes works but speeds continuously decline to
the point of being unusable. I'm not seeing any consistency between what
objects cause what issues and any acls or refresh patterns.
On 10/22/19 3:24 PM, Antonio SJ Musumeci wrote:
> Squid 4.8
> Attempting to get a SMP setup with rock enabled. Instance points to an
> origin web server.
> With "workers 1" everything appears to work fine. If I set "workers 2" I
> get a number of issues:
> If I request a particular object a kid will fail with:
For supported Squid configurations with no other reported errors or
warnings, this assertion indicates a Squid bug. Consider filing a bug
report and posting a stack trace (as well as other relevant reproduction
info) there. Posting an ALL,9 cache.log collected while executing a
single asserting transaction may speed up triage.
> Is there a simple, modern, functional SMP example to compare?
To enable basic SMP features, adding "cache_dir rock..." or "workers N"
(with N greater than 1) is enough.