chn-lee-yumi opened a new issue #6684:
URL: https://github.com/apache/trafficserver/issues/6684
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL abo
masaori335 commented on issue #:
URL: https://github.com/apache/trafficserver/pull/#issuecomment-616367367
I agree with this is an ugly hack:(
I still don't know what is root cause, but this looks LLD specific issue
with debug build. If I use GNU ld instead of LLD, the issue
masaori335 edited a comment on issue #:
URL: https://github.com/apache/trafficserver/pull/#issuecomment-616367367
I agree with this is an ugly hack:(
I still don't know what is root cause, but this looks LLD specific issue
with debug build. If I use GNU ld instead of LLD, th
maskit commented on issue #:
URL: https://github.com/apache/trafficserver/pull/#issuecomment-616421557
I looked into this too. Here's a couple of findings:
- Optimization does some magic. If I add `-O1` to AM_CPPFLAGS on
http2/Makefile.am then link succeeds without the LDADD
traeak commented on issue #6571:
URL: https://github.com/apache/trafficserver/pull/6571#issuecomment-616596029
[approve ci autest]
This is an automated message from the Apache Git Service.
To respond to the message, please lo
traeak opened a new pull request #6685:
URL: https://github.com/apache/trafficserver/pull/6685
Backport PR to make the autests simpler and more resilient.
This is an automated message from the Apache Git Service.
To respond t
shinrich opened a new pull request #6686:
URL: https://github.com/apache/trafficserver/pull/6686
@randall found memory leaks in a previous hostDB fix I pushed
153c08166ed3ba83f92064e607b37210fa34c297.
I spent quite a bit of time last winter tracking down HostDB events that
would tri
randall commented on a change in pull request #6686:
URL: https://github.com/apache/trafficserver/pull/6686#discussion_r411484138
##
File path: iocore/hostdb/HostDB.cc
##
@@ -1420,37 +1418,38 @@ HostDBContinuation::dnsEvent(int event, HostEnt *e)
return EVENT_CONT;
duke8253 commented on issue #6602:
URL: https://github.com/apache/trafficserver/pull/6602#issuecomment-616643314
I'm fine with either, I'll revert back to not changing the names then?
This is an automated message from the Apa
randall commented on issue #6667:
URL: https://github.com/apache/trafficserver/pull/6667#issuecomment-616644622
[approve ci autest]
This is an automated message from the Apache Git Service.
To respond to the message, please l
randall removed a comment on issue #6667:
URL: https://github.com/apache/trafficserver/pull/6667#issuecomment-616644622
[approve ci autest]
This is an automated message from the Apache Git Service.
To respond to the message,
randall removed a comment on issue #6663:
URL: https://github.com/apache/trafficserver/pull/6663#issuecomment-614150227
[approve ci autest]
This is an automated message from the Apache Git Service.
To respond to the message,
zwoop commented on issue #6685:
URL: https://github.com/apache/trafficserver/pull/6685#issuecomment-616660085
Was this cherry-picked with "git cherry-pick -x" ?
This is an automated message from the Apache Git Service.
To res
traeak commented on issue #6685:
URL: https://github.com/apache/trafficserver/pull/6685#issuecomment-616665949
No I didn't do it that way at the time for whatever reason. I can redo this
PR if desired.
This is an automated
sudheerv commented on issue #6602:
URL: https://github.com/apache/trafficserver/pull/6602#issuecomment-616670924
> I don't think we should rename TSContScheduleOnPool() to be
TSContSchedule(). The function requires a pool and it makes more sense to have
the name TSContScheduleOnPool(). I h
sudheerv edited a comment on issue #6602:
URL: https://github.com/apache/trafficserver/pull/6602#issuecomment-616670924
> I don't think we should rename TSContScheduleOnPool() to be
TSContSchedule(). The function requires a pool and it makes more sense to have
the name TSContScheduleOnPool
sudheerv edited a comment on issue #6602:
URL: https://github.com/apache/trafficserver/pull/6602#issuecomment-616670924
> I don't think we should rename TSContScheduleOnPool() to be
TSContSchedule(). The function requires a pool and it makes more sense to have
the name TSContScheduleOnPool
shinrich commented on issue #6686:
URL: https://github.com/apache/trafficserver/pull/6686#issuecomment-616673451
Updated to address @randall's comments
This is an automated message from the Apache Git Service.
To respond to t
SolidWallOfCode commented on issue #6602:
URL: https://github.com/apache/trafficserver/pull/6602#issuecomment-616682942
Given what I suggested [a month
ago](https://github.com/apache/trafficserver/pull/6577#issuecomment-604720341)
I am glad to see @sudheerv and @bryancall have come around
shukitchan commented on issue #6571:
URL: https://github.com/apache/trafficserver/pull/6571#issuecomment-616684827
[approve ci autest]
This is an automated message from the Apache Git Service.
To respond to the message, pleas
shukitchan commented on a change in pull request #6571:
URL: https://github.com/apache/trafficserver/pull/6571#discussion_r411540669
##
File path: plugins/lua/ts_lua.c
##
@@ -26,36 +26,311 @@
#define TS_LUA_MAX_STATE_COUNT 256
+#define TS_LUA_STATS_TIMEOUT 1000 // 1s
Re
shinrich commented on a change in pull request #6645:
URL: https://github.com/apache/trafficserver/pull/6645#discussion_r411545401
##
File path: plugins/experimental/traffic_dump/traffic_dump.cc
##
@@ -597,9 +601,29 @@ global_ssn_handler(TSCont contp, TSEvent event, void
*edat
zwoop opened a new pull request #6687:
URL: https://github.com/apache/trafficserver/pull/6687
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
randall opened a new pull request #6688:
URL: https://github.com/apache/trafficserver/pull/6688
This was introduced in 58b154bb6f253ac8660bd778dd2bf43cb7b60687
This is an automated message from the Apache Git Service.
To resp
shinrich opened a new pull request #6689:
URL: https://github.com/apache/trafficserver/pull/6689
Addresses issue #6300. Details of crash discussed there. We have been
running with this patch since January. Forgot to push this once we verified
its efficacy.
shinrich opened a new pull request #6690:
URL: https://github.com/apache/trafficserver/pull/6690
The current logic will apply the host name and SNI name match check of the
host name would have triggered a SNI policy for verify_client or tls_versions.
After working with this in produc
zwoop commented on issue #6688:
URL: https://github.com/apache/trafficserver/pull/6688#issuecomment-616751195
I assume this is an 8.1.x candidate too ?
This is an automated message from the Apache Git Service.
To respond to t
randall commented on issue #6688:
URL: https://github.com/apache/trafficserver/pull/6688#issuecomment-616753640
> I assume this is an 8.1.x candidate too ?
I think so. It applies cleanly.
This is an automated message f
vmamidi commented on issue #6686:
URL: https://github.com/apache/trafficserver/pull/6686#issuecomment-616753827
If we are trying to fix #5952, I wonder why didn't HttpSM Continuation
cancel pending event when it died? If we think HttpSM canceling the pending DNS
event is working as expecte
randall commented on a change in pull request #6675:
URL: https://github.com/apache/trafficserver/pull/6675#discussion_r411628186
##
File path: tests/gold_tests/bigobj/bigobj.test.py
##
@@ -0,0 +1,129 @@
+'''
+Test client-side communication when GETting a very large object.
+''
traeak opened a new pull request #6691:
URL: https://github.com/apache/trafficserver/pull/6691
This tries to make the slice plugin smarter when origins do bad things to
assets without changing their path including:
* etag change
* last modified change
* content length change
randall commented on issue #6686:
URL: https://github.com/apache/trafficserver/pull/6686#issuecomment-616780380
Running under Instruments.app, I no longer see the HostDB continuation leak.
I do see an assert hit here:
```
Fatal: HttpSM.cc:2880: failed assertion `event == HTTP_TUNN
randall commented on issue #6686:
URL: https://github.com/apache/trafficserver/pull/6686#issuecomment-616780585
@zwoop Can you put this PR on docs and see if the leaks you were seeing go
away?
This is an automated message f
zwoop commented on issue #6658:
URL: https://github.com/apache/trafficserver/pull/6658#issuecomment-616797897
Cherry-picked to v9.0.x branch.
This is an automated message from the Apache Git Service.
To respond to the message
zwoop commented on issue #6688:
URL: https://github.com/apache/trafficserver/pull/6688#issuecomment-616798389
Cherry-picked to v9.0.x branch.
This is an automated message from the Apache Git Service.
To respond to the message
zwoop commented on issue #6688:
URL: https://github.com/apache/trafficserver/pull/6688#issuecomment-616798655
Cherry-picked to 8.1.x
This is an automated message from the Apache Git Service.
To respond to the message, please
zwoop commented on issue #6671:
URL: https://github.com/apache/trafficserver/issues/6671#issuecomment-616828607
is this still an issue with 8.1.x or 9.0.x ?
This is an automated message from the Apache Git Service.
To respond
mlibbey commented on issue #6674:
URL: https://github.com/apache/trafficserver/issues/6674#issuecomment-616865474
Should have noted (thanks @SolidWallOfCode ) in the original description
that range requests that are before the magic offset against good and bad
copies have the same good md5
shinrich commented on issue #6686:
URL: https://github.com/apache/trafficserver/pull/6686#issuecomment-616874375
@randall can you tell what event was delivered to the HttpTunnel that caused
the assert?
This is an automated
shinrich commented on issue #6686:
URL: https://github.com/apache/trafficserver/pull/6686#issuecomment-616874877
@vmamidi The problem isn't that HttpSM died. Rather that multiple events
were created for the HostDBContinutation, and the original logic was not
correctly tracking those event
masaori335 commented on issue #:
URL: https://github.com/apache/trafficserver/pull/#issuecomment-616881628
Thanks for looking.
> Optimization does some magic. If I add -O1 to AM_CPPFLAGS on
http2/Makefile.am then link succeeds without the LDADD change, but not -O0.
Th
masaori335 commented on issue #:
URL: https://github.com/apache/trafficserver/pull/#issuecomment-616890728
I pushed new commit which satisfies both of GNU ld & LLD without duplicated
lines.
This is an automated messa
maskit commented on issue #:
URL: https://github.com/apache/trafficserver/pull/#issuecomment-616900020
> I tried another order, moving libinkevent.a between libhttp2.a and
libhdrs.a. It also works.
This is interesting. You may already have tried this, but adding only a dup
l
masaori335 commented on issue #:
URL: https://github.com/apache/trafficserver/pull/#issuecomment-616931987
[approve ci centos]
This is an automated message from the Apache Git Service.
To respond to the message, pleas
masaori335 opened a new pull request #6692:
URL: https://github.com/apache/trafficserver/pull/6692
I saw below `-Wdeprecated-copy` warning with llvm-10 on macOS.
```
CXX traffic_cache_tool/traffic_cache_tool-CacheTool.o
traffic_cache_tool/CacheTool.cc:259:5: error: definition
masaori335 commented on issue #6692:
URL: https://github.com/apache/trafficserver/pull/6692#issuecomment-616989592
[approve ci autest]
This is an automated message from the Apache Git Service.
To respond to the message, pleas
masaori335 removed a comment on issue #6692:
URL: https://github.com/apache/trafficserver/pull/6692#issuecomment-616989592
[approve ci autest]
This is an automated message from the Apache Git Service.
To respond to the messag
47 matches
Mail list logo
