Hi All,
We think that we have identified a case where the Squid log will show the
transaction result as TCP_HIT even though the cached page was expired.
We generate this condition (under ATS 6.1.1) by doing the following --
* Enable negative caching with 120s time-out
* Request a page from the O.S. via ATS that results in a 404 being
cached
* Disable the connectivity between ATS and its DNS server (TTL is set
for 30s)
* Wait for 120+ seconds to elapse
* Request the same page via ATS again
We observed that the client receives 502 -- Cannot find server. However, the
Squid log shows TCP_HIT/502.
It appears that the Squid log entry is generated as a response to the O.S. DNS
look-up failure without the appropriate via_string is set.
* In HttpTransact::HandleCacheOpenReadHitFreshness()
o The following is set -- SET_VIA_STRING(VIA_DETAIL_CACHE_LOOKUP,
VIA_DETAIL_MISS_EXPIRED)
o However, via_string[VIA_CACHE_RESULT] remains VIA_IN_CACHE_FRESH
* In HttpTransactHeaders::generate_and_set_squid_codes(), the end
result is a TCP_HIT.
1. Does this behavior look like a bug?
2. Any issues or side-effects with doing the following --
In HandleCacheOpenReadHitFreshness(), also SET_VIA_STRING(VIA_CACHE_RESULT,
VIA_IN_CACHE_STALE)
3. In generate_and_set_squid_codes(), the above results in an interim
log_code of SQUID_LOG_TCP_REFRESH_MISS
Then change the error over-rides as follows --
case VIA_ERROR_CONNECTION :
if ((log_code == SQUID_LOG_TCP_MISS) ||
(log_code == SQUID_LOG_TCP_REFRESH_MISS)) <-- add this
log_code = SQUID_LOG_ERROR_CONNECT_FAIL
This results in a final Squid log output of ERR_CONNECT_FAIL/502
Note that we also observed this behavior in a parent-child cache setup, but I
still need to investigate if any additional changes are required to address
that case.
Appreciate any comments on this approach or any better recommendations.
I can create a GitHub PR (ported to master) if there is consensus that this is
a bug and on the approach to address it.
Thanks,
Peter
