/apache/trafficserver/pull/8607__;!!BhdT!lFEKdtxbGBnhcNPCzFPi2u3EzgDOmNN3xGseGw7u7EP6beRGjrNIyDInP4VMtptmCMGtD5cDzstdxXXjFA$
That should essentially be doing what you want for server request pointers as
well.
Can you try this out?
On Tue, Jun 28, 2022 at 2:16 PM CHOU, PETER wrote:
>
> Kit,
&g
g the
coredump problem.
I believe that it should not be too much of a problem in terms of efficiency.
Thanks.
Kit
On Wed, Jun 17, 2020 at 5:11 PM Chou, Peter wrote:
>
> Hi All,
>
> We recently experienced a bunch of assertion failures with the Lua plugin on
> ATS 7.1.4. When the transac
lector reports = " ..
collectgarbage("count") .. " MB")
end
function do_remap()
ts.debug("Running do_remap()...")
end
-Original Message-
From: CHOU, PETER
Sent: Monday, March 14, 2022 10:29 AM
To: dev@trafficserver.apache.org; us...@trafficse
Lua plugin differences between ATS 7.1.x and ATS 9.x.
Is this a lua script you in plugin.config or remap.config ?
Also what method do you use to see if gc is happening or not ?
On Fri, Mar 11, 2022 at 1:22 PM CHOU, PETER wrote:
>
> Hi,
>
>
>
> We have quite a few Lua scripts th
Hi,
We have quite a few Lua scripts that were written for use with ATS 7.1.4 and
earlier. We are looking to migrate to ATS 9.1.1, but it seems like the garbage
collection that is supposed to happen during a "traffic_ctl config reload"
(reload a remap configuration file containing remap lines us
Hi,
Should I open an issue for a fault, such as an assertion failure, that only
happens with binaries compiled with the '-enable-debug' option?
I thought this just affected the compiler optimization options, but perhaps
something logically is being changed?
Although I usually only use this opt
Hi All,
Apologies if this is a duplicate message for anyone. I had to re-subscribe and
re-send as my company had changed my e-mail address.
We are currently working with ATS 9.0.2.
Since the LuaJIT source is no longer included with the ATS source tree, we have
tried using -
1. Default v2.1
Hi,
There is no background thread checking on the original parent. Once the
original parent is marked down, the parent selection process will ignore
it until it becomes retryable after a time-out period.
This is configurable using the "proxy.config.http.parent_proxy.retry_time"
setting.
Thanks,
Hi,
I know there was some discussion on the bug scrub (last week I think) that
enabling debug output in the records.config on any debug tags will greatly
reduce performance.
Is it also the case that using a binary compiled with '-enable-debug' will also
cause performance issues?
Thanks,
Peter
OK, I found the "CONFIG proxy.config.plugin.dynamic_reload_mode INT 0" to
disable this
new feature. After I already modified my plugin ... :-)
From: Chou, Peter
Sent: Friday, March 19, 2021 2:02 PM
To: dev@trafficserver.apache.org
Subject: Help with plugin loading in ATS 7.1.x vs ATS
Hi,
I have a plugin under ATS 7.1.x that performed initialization under the global
plugin context. This initialized data used to be available to the remap
contexts of the same plugin.
I assume that the new remap plugin reloading capability in 9.0.0 means that
this no long works (since remap pl
Regression test -R 2 passed on Ubuntu Linux 18.04 LTS --
REGRESSION_TEST DONE: PASSED
Traffic Server 9.0.1 Mar 2 2021 10:35:45 piplup2
traffic_server: using root directory '/home/pebc/opt/ats-gh-9.0.x'
Apache Traffic Server - traffic_server - 9.0.1 - (build # 030210 on Mar 2 2021
at 10:35:45)
Regression tests -R2 passed with Ubuntu Linux 18.04 LTS 64-bit server. Thanks.
-Original Message-
From: Leif Hedstrom
Sent: Tuesday, December 8, 2020 10:30 AM
To: users
Cc: dev
Subject: [UPDATED: RC1] Re: [E] [VOTE] Release Apache Traffic Server 9.0.0 (RC0)
We’ll cancel the 9.0.0-RC0
+1 for Ubuntu Linux 18.04 LTS 64-bit server with -R2.
From: Bryan Call
Sent: Tuesday, December 1, 2020 1:21 PM
To: annou...@trafficserver.apache.org; Dev ;
Users
Subject: [VOTE] Release Apache Traffic Server 8.1.1 (RC0)
I've prepared a release for 8.1.1. The release notes for 8.1.1 are avail
+1 for Ubuntu Linux 18.04 LTS 64-bit server with -R2.
BTW, is there a sunset date for 7.1.x support?
From: Bryan Call
Sent: Tuesday, December 1, 2020 1:21 PM
To: annou...@trafficserver.apache.org; Dev ;
Users
Subject: [VOTE] Release Apache Traffic Server 7.1.12 (RC0)
I've prepared a release
efine YY_FLEX_SUBMINOR_VERSION 0" was changed
to "4" in the modified version of the file (among other changes).
Thanks.
From: Chou, Peter
Sent: Wednesday, September 2, 2020 3:59 PM
To: dev@trafficserver.apache.org
Subject: Problem with build with 8.1.x?
Hi,
I noticed that after I compiled the 8
Hi,
I noticed that after I compiled the 8.1.x branch that the compilation process
modified a file tracked by git.
This does not seem right to me?
pebc@piplup2:~/git/github/apache-trafficserver$ git status
On branch 8.1.x
Your branch is up to date with 'origin/8.1.x'.
Changes not staged for com
+1 regression tests '-R 2' passed on Ubuntu Linux 18.04 LTS.
-Original Message-
From: Leif Hedstrom
Sent: Monday, August 24, 2020 3:39 PM
To: dev ; users
Subject: [VOTE] Release Apache Traffic Server 8.1.0 (RC0)
I've prepared a release for 8.1.0 (RC0), which is a release focusing on
s
Hi All,
We recently experienced a bunch of assertion failures with the Lua plugin on
ATS 7.1.4. When the transaction is being killed, the Lua plugin calls
ts_lua_destroy_http_ctx() which calls TSHandleMLocRelease() with
http_ctx->server_request_bufp as the buffer argument. The assertion failure
+1
Regression tests '-R2' passed on Ubuntu Linux 18.04 LTS.
* Apache Traffic Server - traffic_server - 8.0.7 - (build # 041413 on Apr 14
2020 at 13:17:45)
* Linux piplup2 4.15.0-96-generic #97-Ubuntu SMP Wed Apr 1 03:25:46 UTC 2020
x86_64 x86_64 x86_64 GNU/Linux
* REGRESSION_TEST DONE: PASSED
+1
Ran regression tests '-R2' on Ubuntu Linux 18.04 LTS.
Apache Traffic Server - traffic_server - 7.1.10 - (build # 041413 on Apr 14
2020 at 13:02:38)
Linux piplup2 4.15.0-96-generic #97-Ubuntu SMP Wed Apr 1 03:25:46 UTC 2020
x86_64 x86_64 x86_64 GNU/Linux
REGRESSION_TEST DONE: PASSED
-Ori
Hi All,
We experienced some core dumps in ATS 7.1.4. My question is whether the pointers
returned by TSHttpTxnServerReqGet() might ever become invalid over the life of
the transaction?
The core dump occurred from the Lua plugin in the
ts_lua_server_request_header_set() function --
if (!http_c
LE_GC64) If it works, we should definitely
document it.
Thanks.
Kit
On Thu, Jun 20, 2019 at 1:35 PM Chou, Peter wrote:
>
> Kit,
>
> This is the only error message seen from diags.log. It did not cause the ATS
> to core dump.
>
> [Jun 20 13:27:45.258] traffic_server NOTE: plu
d be some latency difference between having a small
buffer size vs a larger one (up to some limit), but my guess is it would be
hard to measure because it would be so small.
-Bryan
> On Sep 11, 2019, at 11:50 AM, Chou, Peter wrote:
>
> Hi all,
>
> Sometimes we see lots of EAGAIN
Hi all,
Sometimes we see lots of EAGAIN result codes from ATS trying to write to the
TCP socket file descriptor. I presume this is typically due to congestion or
rate mis-match between client and ATS. Is there any benefit to increasing the
TCP socket buffer size which would reduce the number of
e 19, 2019 11:32 PM
To: dev
Subject: Re: Lua plugin memory limit on Linux.
Thanks for the info.
That's good to know. Have you tried how much more you can allocate?
And when you are over the limit, what is the message you are getting?
Perhaps we can update the doc with this information.
Kit
Kit,
Did not hear back from you regarding the memory limit for the Lua plugin on
Linux.
Previously, I had tested on the command line with the luajit executable, but I
have completed some testing with the ATS master branch.
I am running with Ubuntu 18.04 LTS Server 64-bit which ships with 'LuaJ
Ubuntu 18.04 LTS Server 64-bit
Linux piplup2 4.15.0-47-generic #50-Ubuntu SMP Wed Mar 13 10:44:52 UTC 2019
x86_64 x86_64 x86_64 GNU/Linux
Apache Traffic Server - traffic_server - 8.0.4 - (build # 060321 on Jun 3 2019
at 21:19:31)
HEAD detached at 8.0.4-rc0
Result '-R 2' -- REGRESSION_TEST DO
Regression test -R 2 passed on Ubuntu Linux 18.04 LTS Server 64-bit.
REGRESSION_TEST DONE: PASSED
Apache Traffic Server - traffic_server - 7.1.7 - (build # 060313 on Jun 3 2019
at 13:01:37)
HEAD detached at 7.1.7-rc0
Linux piplup2 4.15.0-47-generic #50-Ubuntu SMP Wed Mar 13 10:44:52 UTC 2019
x8
ce.
Kit
On Wed, May 1, 2019 at 9:09 AM Chou, Peter wrote:
>
> Kit,
>
> Thanks for the response. With regards to the LuaJIT 2.1.0-beta3, I have not
> tried
> using the ts_lua plugin, but with a simple test program using the luajit
> command line I was able to allocate 2.0GB o
aware that luajit-2.1.0-beta3 will have any improvement to
the memory limit problem.
Thanks.
Kit
On Tue, Apr 30, 2019 at 11:05 AM Chou, Peter wrote:
>
> Hi,
>
> Few questions on LuaJIT in relation to the ts_lua plugin.
>
> 1. We did get the following core dump in production.
Hi,
Few questions on LuaJIT in relation to the ts_lua plugin.
1. We did get the following core dump in production. Does this fit the signature
for a LuaJIT memory exhaust?
Program terminated with signal 11, Segmentation fault.
#0 0x2b67b4bfeacd in __run_exit_handlers () from /lib64/libc.so.
Reporting regression tests -R 2 passed on Ubuntu Linux 18.04 LTS Server 64-bit.
-Original Message-
From: Bryan Call
Sent: Monday, March 11, 2019 11:15 AM
To: dev ; users
Subject: [VOTE] Release Apache Traffic Server 8.0.3 (RC0)
I've prepared a release for 8.0.3 (RC0). The release not
Hi,
I get an assertion fault core dump in 7.1.x HEAD from HdrHeap.cc::874. This
seems to be isolated down to commit 739d8d.
What I think is happening is that the copy of the content stored in the cache
uses a different amount of space before and after this change when doing an
un-marshal.
I re-
Regression test -R 2 passed on Ubuntu Linux 18.04 LTS Server.
BTW, built against 7.1.x HEAD matching the git hash below.
The tag did not appear to be in the GitHub repo yet.
-Original Message-
From: Zelkowitz, Evan
Sent: Tuesday, January 22, 2019 3:54 PM
To: dev ; Mailing-List trafficser
Regression test -R 2 passed on Ubuntu Linux 18.04 LTS Server.
-Original Message-
From: Bryan Call
Sent: Tuesday, January 22, 2019 12:48 PM
To: dev ; users
Subject: [VOTE] Release Apache Traffic Server 8.0.2 (RC0)
I've prepared a release for 8.0.2 (RC0). The release notes for 8.0.2 are
+1 Regression Tests passed (-R 2) on Ubuntu Linux 18.04 LTS.
-Original Message-
From: Leif Hedstrom
Sent: Wednesday, November 7, 2018 9:18 PM
To: dev ; Mailing-List trafficserver
Subject: [VOTE] Release Apache Traffic Server 7.1.5 (RC0)
I've prepared a release for 7.1.5 (RC0), which i
Hi All,
I opened an issue on GitHub #3991 a while back. Somebody just responded that
they may be seeing the same thing. Thought I would try the mailing list for
some additional exposure :).
We were running 6.2.1 with RWW=1, and we found that byte range request
responses were truncated. We thou
Compiled on Ubuntu 18.04 LTS. All regression tests passed '-R 1'.
-Original Message-
From: Chris Lemmons [mailto:alfic...@gmail.com]
Sent: Tuesday, July 31, 2018 11:04 AM
To: dev@trafficserver.apache.org
Subject: Re: [VOTE] Release Apache Traffic Server 7.1.4 (RC1)
We're currently runnin
ened a PR #4041 with both memory leaks.
Thanks,
Peter
-Original Message-
From: Chou, Peter
Sent: Thursday, July 26, 2018 1:42 PM
To: dev
Subject: RE: Issue #2416 - Log Collation Memory Leak.
Leif,
1) No problem. It is a two line change so we can carry in our custom 7.1.x if
necessary
8, at 3:14 AM, Chou, Peter wrote:
>
> Hi All,
>
> After some additional study on how the ink_atomic_increment() and
> LogBuffer::destroy() work together in the LogFile::, LogHost::, and
> LogHostList::preproc_and_try_delete() functions, the following is probably a
>
pending logs will be outputed to orphan file,
and them will not be sent to the failover host.
Signed-off-by: Yunkai Zhang
-Bryan
> On Jul 26, 2018, at 2:14 AM, Chou, Peter wrote:
>
> Hi All,
>
> After some additional study on how the ink_atomic_increment() and
led in
LogHostList::preproc_and_try_delete() and destroy is called before the end of
the function (either within the client SM or in the :done part of the function).
Thanks,
Peter
From: Chou, Peter
Sent: Wednesday, July 25, 2018 6:53 PM
To: dev@trafficserver.apache.org
Subject: Issue #2416 -
Hi All,
We are following up on a memory leak that we reported with 6.2.1 and have now
verified to be in 7.1.3 also. What happens is that when Log Collation is
configured and the log collation host is down that the log buffer will be
orphaned and leaked. At first we thought this was a lab only i
Hi all,
I was able to compile the current master branch with a new install of Ubuntu
18.04 Server.
It did not work with 14.04 or 16.04 (unless there is some way to update the C++
compiler).
I did get a diagnostic message during the compile (happened multiple items,
only showing one example).
YvwhbK-KWGqA6Fz3kFJE3ZEZMVvzDDo&s=1eayWyMoU-xaIuSroHrzio06A76Gv1apgRS0OjrTlKQ&e=
>
> Presentations+-+2017?preview=/70255385/74684709/ATSSummit_jemalloc.pptx
>
> Perhaps we can then see what components of the code is contributing to
> the memory buildup.
>
> Thanks.
&
Hi All,
We have been experiencing a slow memory leak with ATS 6.2.1 which requires a
process restart (before memory exhaust) on the order of weeks. We have tried to
gather information with a debug build (compiled with --enable-debug and
--with-tcmalloc-lib), but the process seems to be unstable
Aaron,
Thanks for the response. Jeremy and I were looking for a way to cache an object
that would have been blocked by the cache_promote plugin (which uses the API
function you show below). We are considering adding an over-ride into this
plugin, e.g., special ATS internal header, that could be
Hi,
We see that there are compiler changes on the way for mitigating the
Meltdown/Spectre issues. Is there any recommendation on whether ATS should be
re-compiled, or is ATS alright assuming the built-in Lua JIT is only used for
local, trusted Lua code?
Thanks,
Peter
-24 3:10 GMT+08:00 Chou, Peter :
> Hi,
>
> In adj_system_settings() in proxy/Main.cc, we have (paraphrasing here) --
>
> if throttle > (current - head_room) then
> cur = max = throttle
> setrlimit()...
>
> Would it make more sense to set --
>
>
Hi,
In adj_system_settings() in proxy/Main.cc, we have (paraphrasing here) --
if throttle > (current - head_room) then
cur = max = throttle
setrlimit()...
Would it make more sense to set --
cur = max = throttle + head_room
Otherwise, our users see the warning, try to compensate, an
Hi,
I am taking a look at the collapsed_connection plugin. There appears to be an
assertion failure crash incompatibility between this plugin and changes made in
the 6.2.x branch. TS-4387 appears to require all continuations calling
TSContSchedule() to have a mutex, but in the plugin no mutex i
Hi,
We are looking into some potential memory leaks -- so we are trying to run
traffic_server with the dmalloc library (threaded C++ version libdmallocthcxx)
linked.
I found that the process will abort at start-up inside ink_freelist_new() when
running the new executable.
I found a hint in TS
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
Hi,
I thought the R6.2.1 RC0 passed voting a couple of weeks ago.
Just want to confirm it is still finalized (nothing new pending) and that we're
just waiting on the official release artifacts.
Thanks,
Peter
Hi,
We're wondering if there are plans to release a 6.2.1 version and whether
there's a target date for this?
I looked on the wiki, but there's nothing certain there for point releases.
Thanks,
Peter
Hi,
Isn't there a benefit to 'squash and merge' since it can keep some
work-in-progress commits from ending up in the main branch history? Just
curious if there are other considerations since both squash-and-merge and
rebase-and-merge seem pretty similar (both use fast-forward-merging so no mer
Hi all,
I am working on a fix for TS-4475 which is a core dump in
LogCollationClientSM.cc. I gather that there is not a lot of interest in the
Log Collation client so I am reaching out to the mailing list especially for
folks familiar with inactivity cop.
Background: The core dump occurs when
Hi,
I posted a comment with my RCA for TS-3932 (TOS bits not being set on releases
> 5.3.0) recently.
I have been considering whether this issue can be fixed by moving the call to
set dynamic socket options from Connection::open() to Connection::connect().
Note that the underlying issue is tha
59 matches
Mail list logo
