You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@trafficserver.apache.org by "bneradt (via GitHub)" <gi...@apache.org> on 2023/06/09 21:41:14 UTC
[GitHub] [trafficserver] bneradt opened a new issue, #9821: 10.0.x: OpenReadHead failed for cachekey : vector inconsistency
bneradt opened a new issue, #9821:
URL: https://github.com/apache/trafficserver/issues/9821
When updating one of our internal 9.2.x ATS instances to 10.0.x, I observe the following cachekey notice messages frequently:
```
[Jun 9 19:12:43.041] [ET_NET 25] NOTE: OpenReadHead failed for cachekey D0211905 : vector inconsistency - unmarshalled -1 expecting 4536 in 22712 (base=80, ver=24:2) - vector n=0 size=0first alt=-1[bogus]
[Jun 9 19:13:29.716] [ET_NET 26] NOTE: OpenReadHead failed for cachekey B3C93263 : vector inconsistency - unmarshalled -1 expecting 24096 in 24172 (base=80, ver=24:2) - vector n=0 size=0first alt=-1[bogus]
[Jun 9 19:14:49.779] [ET_NET 29] NOTE: OpenReadHead failed for cachekey 87F9D92A : vector inconsistency - unmarshalled -1 expecting 4224 in 4307 (base=80, ver=24:2) - vector n=0 size=0first alt=-1[bogus]
```
I'm filing an issue for this because I do not see these messages in our 9.2.x instances but I do after upgrading to 10.0.x. As I understand it, there should be no cache compatibility issues with the cache when crossing from 9.x to 10.x.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] 10.0.x: OpenReadHead failed for cachekey : vector inconsistency [trafficserver]
Posted by "bneradt (via GitHub)" <gi...@apache.org>.
bneradt closed issue #9821: 10.0.x: OpenReadHead failed for cachekey <key> : vector inconsistency
URL: https://github.com/apache/trafficserver/issues/9821
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] 10.0.x: OpenReadHead failed for cachekey : vector inconsistency [trafficserver]
Posted by "bneradt (via GitHub)" <gi...@apache.org>.
bneradt commented on issue #9821:
URL: https://github.com/apache/trafficserver/issues/9821#issuecomment-1786309355
Bisecting commits on our prod system, I found the commit that introduces this vector inconsistency behavior:
#9556
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] 10.0.x: OpenReadHead failed for cachekey : vector inconsistency [trafficserver]
Posted by "bneradt (via GitHub)" <gi...@apache.org>.
bneradt commented on issue #9821:
URL: https://github.com/apache/trafficserver/issues/9821#issuecomment-1786348851
I'm not sure what the right direction is here. Thinking about @masaori335 's change further, it is reasonable to avoid downcasting the time_t to uint32_t. Maybe we should just document that the cache is inconsistent in the upgrade from ATS 9 to ATS 10? Clearing the cache resolves the issue. I'll solicit feedback about this so we can figure out whether a cache inconsistency across the major versions is acceptible.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [trafficserver] cheluskin commented on issue #9821: 10.0.x: OpenReadHead failed for cachekey : vector inconsistency
Posted by "cheluskin (via GitHub)" <gi...@apache.org>.
cheluskin commented on issue #9821:
URL: https://github.com/apache/trafficserver/issues/9821#issuecomment-1587805889
The same problem. Clearing the cache doesn't fix it.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] 10.0.x: OpenReadHead failed for cachekey : vector inconsistency [trafficserver]
Posted by "bneradt (via GitHub)" <gi...@apache.org>.
bneradt commented on issue #9821:
URL: https://github.com/apache/trafficserver/issues/9821#issuecomment-1786342935
The type change from uint32_t to time_t changes the type size from 32 bits to 64 bits on 64 bit systems. It makes sense that would change the stored doc structure.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] 10.0.x: OpenReadHead failed for cachekey : vector inconsistency [trafficserver]
Posted by "bneradt (via GitHub)" <gi...@apache.org>.
bneradt commented on issue #9821:
URL: https://github.com/apache/trafficserver/issues/9821#issuecomment-1786329946
> @masaori335 : I wonder whether the issue comes from swapping the order of pinned element here: https://github.com/apache/trafficserver/pull/9556/files#diff-16cd59c47246f90109883f36526055d937b92630ed6e0cdba4664f2328bf1909L301
>
> Do you recall why we switched the order of those elements?
I tested with that order restored, and the box still has cache inconsistency issues. I suppose the issue is rather the change of the type.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
[GitHub] [trafficserver] cmcfarlen commented on issue #9821: 10.0.x: OpenReadHead failed for cachekey : vector inconsistency
Posted by "cmcfarlen (via GitHub)" <gi...@apache.org>.
cmcfarlen commented on issue #9821:
URL: https://github.com/apache/trafficserver/issues/9821#issuecomment-1588218775
The `-1` return is likely coming from `HTTPInfo::get_handle(char *buf, int len)`
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] 10.0.x: OpenReadHead failed for cachekey : vector inconsistency [trafficserver]
Posted by "masaori335 (via GitHub)" <gi...@apache.org>.
masaori335 commented on issue #9821:
URL: https://github.com/apache/trafficserver/issues/9821#issuecomment-1786448680
Let's revert the change made by #9556. I didn't noticed it breaks cache compatibility when I fixed coverity issue. I think using `uint32_t` is enough for cache and cache compatibility is more important.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
Re: [I] 10.0.x: OpenReadHead failed for cachekey : vector inconsistency [trafficserver]
Posted by "bneradt (via GitHub)" <gi...@apache.org>.
bneradt commented on issue #9821:
URL: https://github.com/apache/trafficserver/issues/9821#issuecomment-1786313138
@masaori335 : I wonder whether the issue comes from swapping the order of pinned element here:
https://github.com/apache/trafficserver/pull/9556/files#diff-16cd59c47246f90109883f36526055d937b92630ed6e0cdba4664f2328bf1909L301
Do you recall why we switched the order of those elements?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: issues-unsubscribe@trafficserver.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org