S, we are still seeking the
way for CLFUS. Anyway, let us make sure we can lock the memory down.
thanks
在 2012-12-09日的 00:47 +0800，ming@gmail.com写道：
> laster week, we have submited the patches for TS-1006,
> https://issues.apache.org/jira/browse/TS-1006 .
>
> the memory issue is
hmm, change in the interface, do you have tested the SplitDNS?
在 2012-12-10一的 11:12 -0600，Alan M. Carroll写道：
> Although we try to keep fixes separate, the fix for these two bugs ended up
> being too intertwined to readily separate.
>
> The primary change of interest is to split the HostDB in to
laster week, we have submited the patches for TS-1006,
https://issues.apache.org/jira/browse/TS-1006 .
the memory issue is the most critical issue here as we are running TS in
heavy duty, here is the 2nd version of our memory patch in TS-1006:
https://issues.apache.org/jira/secure/attachment/125
+1
在 2012-12-05三的 16:11 -0700，Leif Hedstrom写道：
> I'd like to suggest we use a second IRC channel (#traffic-server-ci) for all
> build-bot and other automated Continuous Integration notifications. This has
> two benefits (for me at least):
>
> 1) I can easily see / scroll back / archive the CI
personally, I am trying to make the master tree as clean as possible, a
clean tree is very important for new comers, when they can feel happy
with the way we pushing the project, I'd like every one love this
project, for using and coding everyday.
Well, we make big changes when it really ought to,
hmm, I have put him in the author of git commit:
commit 6dadea4cea3da45e432dc52c7f9dadbf891b510a
Author: Craig Forbes
Date: Fri Nov 9 00:44:52 2012 +0800
TS-1564: Logs created via the API in plugins are not rolled
should we strict on the style? which one is prefer then?
在 2012-11-09五的 0
+1
在 2012-10-16二的 15:24 +0200，Daniel Gruno写道：
> Hi fellow dev@ people,
>
> We've been discussing the possibility of adding a page where third
> parties can get a mention if they support or do consulting on Apache
> Traffic Server, in the hopes that people seeking professional help can
> find it.
v3.2 have some broken Range codes changed by amc, which is fixed in the
master later. in our own 3.2 tree, we have reverted the changes in
Range, TS-475 TS-1265
FYI
在 2012-10-12五的 15:10 -0600，Leif Hedstrom写道：
> On 10/12/12 3:07 PM, Eagen, Dave wrote:
> > Yes, 3.2.3 crashed when we tried it with l
my buildbot get the following fail when I building the 3.2.x tree:
> cc1plus: warnings being treated as errors
> Vec.h: In function ‘int main(int, char**)’:
> Vec.h:616: error: assuming signed overflow does not occur when assuming that
> (X + c) < X is always false
> Vec.h:616: error: assuming
I will start to merge all the patches in cluster, discard the current
git master in a not so stable status.
thanks
--zym
yeah, glad to hear that you guys may take this hard task, and my team
may help on review & guide you on the cache rebuild. as it will be a
tough task, and the redesign of the very low disk layout may introduce
big problem, I'd suggest you post your desgin docs on the cwiki:
https://cwiki.apache.or
jtest is tool for client, maybe someone would like to run on Windows, I
don't think we should not support win32 here.
在 2012-05-13日的 00:18 +，Igor Galić写道：
> And here's the same for win32
>
> ---8<
> mgmt/cluster/VMap.cc:1141: // Now for WIN32, we have to find the nte_context
> for the
aha, I miss that DLL related codes, I will make another patch.
在 2012-05-13日的 00:02 +，Igor Galić写道：
>
> - Original Message -
> >
> >
> > - Original Message -
> > > Updated Branches:
> > > refs/heads/master a79a638d8 -> 53ebc2f72
> > >
> > >
> > > TS-672 cleanup Win32 refe
patch submitted on TS-672, I think that is safe to commit it at any
time, there is nothing we need to test as all lines removed is just
#ifdef out.
在 2012-05-10四的 14:57 +，Igor Galić写道：
>
> - Original Message -
> > AFAIK, we do not have any plan on supporting WIN32, and some of the
>
AFAIK, we do not have any plan on supporting WIN32, and some of the
codes for WIN32 platform is deleted in the past, but we have many random
pieces still in the core. should we nuke them out completely? It would
be a good point as we are going to do a V3.2 release.
+1 from me
thanks
--
zym, Zha
the alarm function & setting still valid in the git master, I think that
maybe a good solution for some process monitor requirement. and
example_alarm_bin.sh can help you send out mail when traffic_server die
or restarted.
can we include the example_alarm_bin.sh in the make install? we should
put
YES, we do have some users interest in the traffic_shell, I think we can
hold back for sometime. :D
+ 1 for leave it there
在 2012-04-04三的 20:44 -0700，James Peach写道：
> Hi all,
>
> In various IRC conversations, the opinion has been expressed that no-one uses
> traffic_shell and that it should be
I think that is a event which is cancelled and destructed, but still
running in other thread, the event id in our stack show the same issue
as TS-1114, with turnout to be a locking issue, where we should protect
all the vol open/write.
the event is free and the same place is filled with other data
Alan:
I think we have make many effort on this issue already, and fairly
sure that we are closing to the root cause of it. From my team side,
we'd like to keep tracking it instead of hiding the issue. please
holding back the commit, and give us more detail on the crashing back
traces as much as p
thanks Otto, I have introduced your plugin in the Chinese channel.
cheers
在 2012-03-07三的 15:28 +0100，Otto van der Schaaf写道：
> Hi All,
>
>
> I closed the old jira ticket for the gzip plugin, and created a new
> one with the source code of our compressor code attached.
> https://issues.apache.or
hi:
We have submit the patch for TS-1114, for master. After that fix, our
TS can run very stable in production. please help review and test if you
have any crashing issues.
the fixs that will improve stability include the following commit
(already in master):
TS-
TS-1109
Hopes that will
it turns out if you need to get a full core dump when you testing TS,
disable the internal stack dump will help, otherwise may miss a core
file when TS crashed.
CONFIG proxy.config.stack_dump_enabled INT 0
all developers should make this change in the records.config.
the root cause if the sign
when I am testing the 3.0.3 release, it turns out that my gentoo system
will put an incorrect interface name for cluster "CONFIG
proxy.config.cluster.ethernet_interface STRING eth0:", that will stop
the manager process from starting, it is a small case for me, but may
not for other guys that is new
besides DNS, UDP is|maybe needed in the following system:
cluster:
I am not sure if we can handle >2Gbps inter cluster tcp traffic,
maybe we need to think of the UDP implements of the intercluster
communication.
ICP:
it is udp based.
prefetch:
/* This lock fails quite often. This ca
hi:
When checking the request to Origin Server, we find that some of the
URL will show up multiple times in our log, and we dump the cache info:
http://people.apache.org/~zym/dump-190.2.html
dumps a static jpg file, with 2 alternates
http://people.apache.org/~zym/dump-190.3.html
dumps a static
we know that Vol locking is the most expensive lock in cache layer, will
we bind Vol and CPU(or thread) a good idea?
--
zym, Zhao Yongming.
aka: yonghao @ taobao.com
+1
tested on Gentoo x86_64, checksum ok.
在 2011-12-04日的 22:28 +，Igor Galić写道：
> Hi folks,
>
> I've prepared a package for a v3.0.2 release. Please take a look at the
> artifacts, check the STATUS/README/CHANGES files, and do builds and tests.
> After finishing your examination of the relea
+1
we use git in every day of TS, let us move forward.
在 2011-12-01四的 17:00 -0700，Leif Hedstrom写道：
> On 12/1/11 4:56 PM, Leif Hedstrom wrote:
> > On 11/29/11 12:04 PM, Bryan Call wrote:
> >> +1 git
> >>
> >> I use git for all my work and am accustomed to cloning svn repos into
> >> git. There a
+1
taorui and me will focus on landing TS-857, that will help us do more
testing.
在 2011-11-22二的 01:37 +，Igor Galić写道：
>
> - Original Message -
> > +1 (binding)
>
>
> Soon as svn.eu mirror is synced up after that massive ooo commit,
> I'll do the merging of the currently voted prop
+1
the fewer options the better.
在 2011-10-15六的 09:20 -0600，Leif Hedstrom写道：
> As per the discussions and patch provided on
> https://issues.apache.org/jira/browse/TS-973, I'd like to eliminate this
> configuration entirely:
>
> CONFIG proxy.config.http.verbose_via_str INT 2
>
>
> Inste
when browsering the codes in prefetch:
> /* skip if it is encoded */
> c_type = resp->value_get(MIME_FIELD_CONTENT_ENCODING,
> MIME_LEN_CONTENT_ENCODING, &c_type_len);
> if (c_type) {
> char type[64];
> memcpy(type, c_type, c_type_len);
> type[c_type_len] = 0;
> Debug("Prefetc
as we know, we have some users need more control over the content and
would like to speed up whole process when TS going to origin servers, we
I exploring the prefetch codes, something really intersting comes up:
the prefetch is designed to parse the html pages and get all html
resources out then
in my experience with --enable-debug, that is a supper of the default
config, and it should rise more issues. as the build box just do the
basic building and regression test, get it a little strict is a good
idea.
在 2011-09-09五的 08:06 +，Igor Galić写道：
> * change all of current buildbots to --en
we have fixed the problem in trunk, please test with latest trunk and
report back if it is fixed in you situation.
be aware, that v3.0.1 is not fixed too, you should test the svn trunk.
you may get the information from:
https://issues.apache.org/jira/browse/TS-880
thanks
在 2011-08-22一的 09:48 +0
to me sounds like something mixed with root&user permissions.
在 2011-07-22五的 22:25 -0700，Alan D.Cabrera写道：
> SSLCertLookup.cc:236: fatal error: opening dependency
> file .deps/SSLCertLookup.Tpo: Permission denied
this is the problem of cluster communication interface, it should be set
to the existing real interface or loopback interface, change
records.config to:
CONFIG proxy.config.cluster.ethernet_interface STRING lo0
this case often rise up when you install a binary package, and where the
default real i
+1
tested on gentoo & rhel 5.4 64bit
在 2011-07-16六的 00:06 -0400，Igor Brezac写道：
> On Sat, Jul 16, 2011 at 12:03 AM, Leif Hedstrom wrote:
> > On 07/15/2011 05:12 PM, Igor Galić wrote:
> >>
> >> Thanks to Bryan for pointing this out the blindingly obvious number
> >>
> >> I fixed that and rolled t
I am not sure the information in send to Jui-Nan Lin, CC'd here.
users report that trafficserver can not run with port=80, in FreeBSD
with current ports. I think that is the reason we do not run
traffic_server directly from init. it is traffic_manager's duty to open
files and sockets.
please help
asked by Jan-Frode, I think that the change maybe take on our
installation layout. should we do it? if so, I will file the bug.
thanks
--
zym, Zhao Yongming.
aka: yonghao @ taobao.com
I think I need to comment on the status:
we have about 7 crashes during our pre production simulation testing,
that is testing with production hardware/software etc
hard & soft:
Dell DCS5120， 1 X L3406 @ 2.27GHz， 16G ram, 1 X 160 SSD + 3 X 600G
SAS, 2 * GBE with bonding
RHEL 5.4 x86_64. TS ve
cast my vote here:
* Logs: Restore old log buffer size.
Trunk patch: http://svn.apache.org/viewvc?rev=1131473&view=rev
Jira: https://issues.apache.org/jira/browse/TS-820
+1
* SOCKS: Assertion/abort when starting TS with SOCKS proxy enabled
Trunk patch: http://svn.apache.org/vi
you should send it in dev@.
the short answer is YES we do.
TrafficServer before opensource have a module to convert origin FTP
protocal to HTTP protocal, and then serving http content to users.
Leif have nuked the last codes in core one year ago, you may get info
from:
> zym@zym6400 ~/workspace
+1 for release it.
btw, I am going to put v2.1.9 in production and have spread out that
v3.0 will be release in 2weeks, hopes not make any trouble
:D
在 2011-05-31二的 14:13 -0600，Leif Hedstrom写道：
> Hi all,
>
> I'd like to start a discussion on the viability of a 3.0.0 release in
> ~1-2 weeks. I
amc:
maybe the ink_res_sockaddr_union is my fault :D just to get things fix,
but no other long term planning. please refer to TS-313.
I think the codes is hard to adapt for IPV6, I may agree with John that
we need to do something more to get it done, feel free to change.
dns is my lovely place,
look some more technical detail needed, you may get some basic http
cache background information from the SDK guide.
from that Admin Guide, just remember:
User URL -> Maped URL -> MD5 key -> Disk
from the SDK guide and codes, you man find out the very down side
storage interface, which is far com
please refer to amc's project
http://people.apache.org/~amc/tiphares/home.html
be sure to stick with the latest unstable release or svn trunk.
you may need to setup the forward proxy before testing transparent:
http://trafficserver.apache.org/docs/v2/admin/trouble.htm#enable_forward_proxy
PS:
Tr
b 28, 2011 at 9:49 AM, Leif Hedstrom wrote:
> > On 02/28/2011 02:25 AM, ming@gmail.com wrote:
> >>
> >> hi, all:
> >> We have now install plugins into libexecdir, which is recommend for
> >> used for executable files in most distributions, such
The Regression tests is for default directory layout, any change in
directories, may expect fail.
在 2011-03-01二的 01:41 +，Igor Galić写道：
>
> - Original Message -
> > Hi all,
> >
> > I've prepared a package for a v2.1.6 release. Please take a look at
> > the artifacts, check the STATUS/
hi, all:
We have now install plugins into libexecdir, which is recommend for
used for executable files in most distributions, such as fedora:
lib/ contains object files and libraries that are not designed to be
directly utilized by users or shell scripts. The libexec/ directory
contains small hel
from what I can guess, TS' read-while-writer is a connection-collapsing
like implement. TS have a http_tunnel that will make use of the same
data for different output, maybe that is the magic point, right?
After talk with Leif, I have the following config setup:
CONFIG proxy.config.cache.enable_re
can we include TS-597 https://issues.apache.org/jira/browse/TS-597 ?
this fix will cleanup the fallback tricky in SplitDNS, make sure
SplitDNS and DNS works at any config env.
thanks
在 2010-12-19日的 21:52 -0700，Leif Hedstrom写道：
> Hi all,
>
> I'd like to suggest that we aim for a v2.1.5 release d
we are thinking of L7 service health checking in connection manager to
OS, still investing.
if you want do full balance, that is hard, I think we have many
checkpoints:
1, HostDB strict RR, proxy.config.hostdb.strict_round_robin, that may
make all HostDB response strict RR.
2, need a strict RR con
Hi,
the codes have many reference to www.inktomi.com
( http://pastebin.com/wGsUSExZ ), even testing in the DNS module, that
may make some trouble as www.inktomi.com had gone for years:
z...@zym6400 ~/workspace/trafficserver $ host www.inktomi.com
www.inktomi.com is an alias for yst-web1.yahoo.com
the common forwarding proxy should at least set:
* traffic_line -s proxy.config.reverse_proxy.enabled -v 0
* traffic_line -s proxy.config.url_remap.remap_required -v 0
have you change proxy.config.url_remap.remap_required to '0'? the
default is '1' for reverse proxy.
在 2010-11-23二的 12:42
that is one ugly thing, changed to Apache-2.0. :D
please aware that the layout config maybe need some tweak, please tell
me what your concern.
thanks.
在 2010-11-18四的 14:26 +0100，Mladen Turk写道：
> On 11/18/2010 01:09 PM, ming....@gmail.com wrote:
> > hi, all:
> >per jMCg, it
hi, all:
per jMCg, it is a start even it is ugly, so here is our SPEC file for
trafficserver on redhat, it works for our half-production env.
http://zymlinux.net/trafficserver/rpm/
have some Chinese on the page, should not be a problem to find out the
file to download.
if you have any questio
guys, I want to make more DNS and HostDB change:
https://cwiki.apache.org/confluence/display/TS/HostDBandDNS
the basic idea is, we make HostDB call DNS, move SplitDNS under DNS
only, give DNS a more flex Handler logical. after then, we'll make API
and Hooking enhangcement etc.
any idea?
--
--
Great if you can help consider how to do localized document management,
we are close to finish the Admin Guide translating to Chinese, we don't
want to manage a outdated translation anyway :D
thanks
在 2010-10-23六的 01:08 +，Igor Galić写道：
> Hi folks,
>
> In the next two weeks I'll try to migra
seems you are based in China, if you don't mind, I can call you.
please mail me offlist.
在 2010-06-29二的 15:28 +0800，Reynold Chen写道：
> Hi Friends,
>Does anyone have time to help me out about "case
> HttpTransact::ORIGIN_SERVER_OPEN in set_next_state() in HttpSM.cc"?
>
>
> Best regards,
>
59 matches
Mail list logo
