fre 2006-09-01 klockan 18:00 -0300 skrev Gonzalo Arana:
You may want to check this:
http://webs.sinectis.com.ar/garana/patches/squid/1-add-gz.patch.dpatch.CVS
Unless I am mistaken, that is the latest patch I've used for content
compression.
I will dig into this during the weekend and
On Thu, 2006-08-31 at 09:50 +0200, Guido Serassio wrote:
VisualStudio and MinGW share the same compatibility problems: IPC, fd
= socket stuff, etc.
I know.
But VisualStudio has two main pros:
- It's a well known (the best known I think) development environment
for Windows (here I'm
ons 2006-08-30 klockan 10:38 +0200 skrev Guido Serassio:
I think that we could try to add MinGW to the Squid natively
supported platforms, leaving only the VisualStudio support into the nt branch.
+ here. I'd love to se MinGW as an official platform.
Btw, what do you see as the main benefits
sön 2006-08-27 klockan 20:21 +1000 skrev Robert Collins:
Or we could require openSSL 9.8.x?
For that part of the code yes. It's not normally needed, and it's fine
if it's not compiled for all OpenSSL versions.
These lines of code exists mainly for being able to decrypt the SSL
traffic using
sön 2006-08-27 klockan 20:48 +0200 skrev Guido Serassio:
I have published a patch with a configure check into bug #1716:
http://www.squid-cache.org/bugs/attachment.cgi?id=1198.
What is your opinion ?
see bug.
Regards
Henrik
signature.asc
Description: Detta är en digitalt signerad
Looks quite nice I think.
And while I agree with Adrian that perhaps we should try to not do
additional refactoring and API changes in Squid-3.0 I do not think this
kind of things should be sitting in a private branch. If it's finished
and looking good but not a candidate for 3.0 then it's a good
Why is HttpReply/Request (HttpMsg subclasses) using manual reference
counting by HTTPMSGLOCK/UNLOCK macros instead of automatic reference
counting by RefCount?
I find this design quite error prone, especially considering that
several of the LOCK/UNLOCK pairs is crossing code boundaries. One
lör 2006-08-26 klockan 17:58 +0200 skrev Guido Serassio:
For testing, I have replaced PEM_ASN1_write() call with the
PEM_write_SSL_SESSION() macro, and this seems to fix all the errors
that I have listed in bug #1716.
What is the history of this comment ?
It depends on the version of
tor 2006-08-24 klockan 10:27 +0800 skrev Adrian Chadd:
Mark from Wikimedia has popped up and noted that there's been some
issues with the Squid-2.6 deployment. Specifically, people are
being prompted to download pages rather than display them.
It only started after they installed Squid-2.6.
tor 2006-08-24 klockan 15:14 +0800 skrev Adrian Chadd:
Well, whats changed between squid-2.5 and squid-2.6 which may play a part?
The major change related to thisis that we now server side support
content negotiation using ETag and If-None-Match to find which cached
entity variant (identity vs
Squid-2.6 looks pretty decent at the moment with no major issues known.
The only larger bug scheduled to be fixed is delay pool fairness where
connections currently can get completely starved when throttled by
shared pools.
Apart from that Squid-2 maintenance i now switching to pure bugfixing
fre 2006-08-25 klockan 07:07 +1000 skrev Robert Collins:
- Add to ERR_INVALID_REQ error file User: %aP
- reload error messages
ERR_INVALID_URL crash also confirmed.
Whats strange is that we're seeing error pages with no requests.
client_side is setup to create an empty request if
fre 2006-08-25 klockan 08:07 +1000 skrev Robert Collins:
Ok. this is another thing to forward port then - 3.0 still makes a stub
request object (which is yucky).
If it does then thats something new. In Squid-2 the request_t is created
by urlParse()...
The client-side request exists earlier,
tor 2006-08-24 klockan 09:35 -0300 skrev Giancarlo Razzolini:
I can be talking a lot of s... but, i remember seeing something about
gzip compression being broken for some sites on internet explorer, after
microsoft released the monthly patches.
Brokenness exists at many fronts, both servers
From time to time I and a few other developers gets more requests than
we can handle, and I wonder if there is interest in setting up an
internal mailing list where we can ask if there is another squid
developer who can take the assignment.
If there is interest in being in a such mailinglist
On Wed, 2006-08-23 at 11:52 +0200, Kinkie wrote:
Yes.. can't do much about it currently. I will probably have to write a
custom macro. At this time two clicks are required, unfortunately.
The problem is not the two clicks, it's two searches to find the correct
entry in long lists of questions
On Mon, 2006-08-21 at 22:17 +1000, Robert Collins wrote:
I'd like to make the shutdown procedure async as well - allowing things
like swap log writing to use calls elsewhere in the codebase that are
async - like the disk engines.
Makes sense.
Heres my proposed api:
- on EventLoop you can
Was reading the new logfile stuff, and logfileClose() in combination
with logfileFlush() looks dangerous. To me it looks like the last log
segment is lost.
Regards
Henrik
signature.asc
Description: Detta är en digitalt signerad meddelandedel
mån 2006-08-21 klockan 12:50 +0200 skrev Henrik Nordstrom:
Was reading the new logfile stuff, and logfileClose() in combination
with logfileFlush() looks dangerous. To me it looks like the last log
segment is lost.
Nevermind that. Was mixing up the logfileFlush and logfileFlushEvent
functions
mån 2006-08-21 klockan 17:23 +0200 skrev Kinkie:
This said, I'm not opposing the idea; I'd just like to get more reasons
to do it :)
The main reason is that the FAQ is not a document, it's a collection of
very many small notes, loosely coupled together in sections to try to
make some order out
lör 2006-08-19 klockan 04:18 -0600 skrev Adrian Chadd:
Its just a malloc() and then sprintf().
Feel free to do what you think you have to, if you haven't already.
Perhaps replace it with a memBufPrintf to be consistent with the rest of
the code?
Regards
Henrik
signature.asc
Description:
fre 2006-08-18 klockan 02:01 +0200 skrev Henrik Nordstrom:
Ideas on how to avoid starvation when there is multiple connections
competing for a delay pool in the generic comm loop framework is
welcome.
Currently the comm loops take an overly simplistic view of delay pools,
and simply kicks
2.6.STABLE3 has now been released.
Regards
Henrik
signature.asc
Description: Detta är en digitalt signerad meddelandedel
On Thu, 2006-08-17 at 01:07 +0200, Kinkie wrote:
On Thu, 2006-08-17 at 00:03 +0200, Henrik Nordstrom wrote:
ons 2006-08-16 klockan 01:36 +0200 skrev Kinkie:
Hi all,
this is a second attempt at getting autoconf/automake to
understand
our desire to get builds faster, and this attempt
Ideas on how to avoid starvation when there is multiple connections
competing for a delay pool in the generic comm loop framework is
welcome.
Currently the comm loops take an overly simplistic view of delay pools,
and simply kicks all deferred connections alive once per second to have
them react
ons 2006-08-16 klockan 18:41 +0300 skrev Tsachi:
Looking at netstat it shows that squid hold many open sockets (even
180 and more) only for one client.
You should not see this with 2.6.
I would like to know if anyone had a chance to check a similar
scenario to this with version 2.6.
My
ons 2006-08-16 klockan 01:36 +0200 skrev Kinkie:
Hi all,
this is a second attempt at getting autoconf/automake to understand
our desire to get builds faster, and this attempt works for me.
What I don't quite get is why we should need to have configure look for
ccache? It's trivial to do this
On Mon, 2006-08-14 at 10:32 +0200, Guido Serassio wrote:
err-request = requestLink(request);
But not always err-request is linked to a request, it could be a problem ?
You have found places where err-request is assigned without a link?
-src_addr
Some confusion here, sometimes:
tis 2006-08-15 klockan 20:12 +0200 skrev Pawel Worach:
Henrik pointed out on IRC that the changeset at
http://www.squid-cache.org/Versions/v2/2.6/changesets/10924.patch
doesn't actually work, that is because debug() can't be used that
early.
Applied.
signature.asc
Description: Detta är
tis 2006-08-15 klockan 22:46 +0200 skrev Guido Serassio:
In client_side.c at line 3932, 3943, 4073 and 4285 (not used code) I
can't see any err-request assignment.
All of these is before the request has been parsed, so there is no
request...
So like this ?
err-auth_user_request =
Think we are more or less set now for a stable3 release.
Please test the tree some extra. And if you know any issues you think
should be fixed for stable3 then now is a goot time to speak up.
Regards
Henrik
signature.asc
Description: Detta är en digitalt signerad meddelandedel
sön 2006-08-13 klockan 10:27 +0800 skrev Adrian Chadd:
On Sun, Aug 13, 2006, Henrik Nordstrom wrote:
/home/henrik/SRC/squid/commit-2.6/src/comm.c: In function
'CommWriteStateCallbackAndFree':
/home/henrik/SRC/squid/commit-2.6/src/comm.c:85: warning: comparison
between pointer
sön 2006-08-13 klockan 22:37 +0200 skrev Guido Serassio:
Hi,
I'm trying to close the Bug #212.
Someone could review the proposed patch ?
It' should be incomplete, some errors still need to be verified.
Looked at it and my gut feeling is that errorCon should be extended with
a request_t
lör 2006-08-12 klockan 09:51 +1000 skrev Robert Collins:
That is - isn't the for loop completely irrelevant ?
It is.. just a matter of programming style (complex if condition, vs
smaller conditions with break).
Regards
Henrik
signature.asc
Description: Detta är en digitalt signerad
lör 2006-08-12 klockan 11:22 +0200 skrev Henrik Nordstrom:
lör 2006-08-12 klockan 09:51 +1000 skrev Robert Collins:
That is - isn't the for loop completely irrelevant ?
It is.. just a matter of programming style (complex if condition, vs
smaller conditions with break).
This said I agree
On Wed, 2006-08-09 at 13:35 +0300, Liudas Bakšys wrote:
http_port 192.168.0.254:8080 transparent vhost
Don't use vhost here... vhost is for accelerator mode, not Internet
proxy mode. transparent is all you need for a transparently intercepting
Internet proxy.
In accelerator mode CONNECT
tis 2006-08-08 klockan 07:51 +0800 skrev Steven:
I did have one other thought. If you set max-object-size on a cache_dir,
it effectively allows an object to use up to max-object-size woth of RAM
before it will start to swap out. I can understand this being needed if
there is no
tis 2006-08-08 klockan 17:29 +0800 skrev Adrian Chadd:
* Reimplement logfile-daemon.c to not use stdio. I've noticed via strace
that the Linux stdio implements its reading using one-byte read()s.
That depends on your stdio buffer setting. To get bigger reads you need
block/full buffering on
mån 2006-08-07 klockan 17:50 +0800 skrev Steven:
Once an object reachs store_maxobjsize (in my case 128K), it falls through
and eventually runs storeCheckCachable() and figures out that the object
is not cachable. The problem here is that if all data has been sent to
the client, there is
mån 2006-08-07 klockan 18:26 +0800 skrev Steven:
2006/08/04 16:01:42| WARNING! Your cache is running out of filedescriptors
2006/08/04 16:01:42| comm_call_handlers(): WARNING defer handler for
fd=12 (desc=HTTP Socket) does not call commDeferFD() - backing off
manually
Thinking.. what is the
mån 2006-08-07 klockan 13:04 +0200 skrev Guido Serassio:
Yes, I also prefer a solution without any special markup, but this
solution warrants that the commiter must know what is doing, and it's
a not so big effort like the old 2.5 patch management :-)
What about the following simple
mån 2006-08-07 klockan 19:24 +0800 skrev Steven:
If you let the defer function call commDeferFD, you don't run through the
comm loop 2 times to cause a FD to back off (once to set the defer/backoff
flag, and once to actually back off). The code in the comm loop is there
as a fail-safe, but
mån 2006-08-07 klockan 19:38 +0800 skrev Steven:
From what I can tell, the attached patch will break the current
behaviour. storeCheckCachable() needs to remove the ENTRY_CACHABLE flag
so that other parts of the code know that they can free the memory.
There is only two users of
Steven, do you think this could be the cause of Bug #1304?
Never got to the bottom with that bug and it seemed to disappear, but it
feels like this could be the root cause...
Regards
Henrik
signature.asc
Description: Detta är en digitalt signerad meddelandedel
mån 2006-08-07 klockan 14:44 +0200 skrev Guido Serassio:
What about an optional tag allowing a minimum length override ?
If we go to the tagged/guided form there will most likely be no lower
limits other than there must be a title and a following description.
For now I thinkt I'll just ignore
mån 2006-08-07 klockan 17:50 +0800 skrev Steven:
Once an object reachs store_maxobjsize (in my case 128K), it falls through
and eventually runs storeCheckCachable() and figures out that the object
is not cachable. The problem here is that if all data has been sent to
the client, there is
lör 2006-07-29 klockan 09:35 +0200 skrev Guido Serassio:
I have found an occasional DNS resolution error when browsing
www.microsoft.com.
I have seen the error only few times, less then 10, but the odd thing
is that this happens always only with www.microsoft.com and sometime
it was
lör 2006-08-05 klockan 15:49 +0800 skrev Adrian Chadd:
When I went through this exercise a while ago my solution was to make sure
I 'kicked' (ie, started another comm read event) the server side if
all the clients had read all the data they could (and thus -someone- had to
kickstart an IO
lör 2006-08-05 klockan 15:45 +0800 skrev Steven:
if (fd = 0 mem-inmem_hi - mem-inmem_lo SM_PAGE_SIZE +
Config.Store.maxInMemObjSize + Config.readAheadGap) {
- storeDeferRead(e, fd);
- return 1;
+ if(storeLowestMemReaderOffset(e) != mem-inmem_hi) {
+
On Tue, 2006-08-01 at 01:53 +1200, Reuben Farrelly wrote:
I wrote this a while ago...does it seem suitable to put in -CVS or is it a
bit long?
Length is OK I think, but Adrian is the guy to speak to for COSS
content. COSS is still experimental so there is no feature freeze in the
COSS
mån 2006-07-31 klockan 11:25 +0200 skrev Eduard Veleba:
the patch we sent you contains another one small change - possibility of
setting
expiration time in refresh_pattern in seconds, not only minutes. It
simply doesn't
multiply the time with 60, when there's letter 's' right after the
mån 2006-07-31 klockan 12:43 +0200 skrev Guido Serassio:
Hi Henrik,
At 12.39 31/07/2006, Henrik Nordstrom wrote:
Hmm.. ah. cvsps got a bit confused. Not by this change but by the tag
being moved after the changesets had been updated (minor packaging
bugfix, had forgot to update ChangeLog
mån 2006-07-31 klockan 21:49 +0800 skrev Adrian Chadd:
On Mon, Jul 31, 2006, Steven Wilton wrote:
The code does only touch the coss directory (except for the comments in
cf.data.pre) if that makes any difference.
Or maybe I should check whether it's been released before posting :)
lör 2006-07-29 klockan 19:45 +0200 skrev Guido Serassio:
I want to test the forward port of HTCP changes from 2.6 to 3.0, but
I don't know how to check al the HTCP functionality.
Most of the code is tested by peering with 2.5 and 2.6 in both
directions.
The CLR support need a HTCP client.
sön 2006-07-30 klockan 09:56 +0200 skrev Guido Serassio:
I think that likely you have found the problem
look the log attached
to this Squid bug on Debian opened by the same people:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=380020
2006/07/26 16:52:21| ipcCreate:
sön 2006-07-30 klockan 19:12 +0200 skrev Guido Serassio:
Any suggestion on the peering setup ?, excluding the htcp option ... :-)
Nothing else particular I can think of..
The CLR support need a HTCP client. Unfortunately we don't have any..
Perhaps one should be written for HTCP and ICP.
tor 2006-07-27 klockan 16:00 +0800 skrev Steven Wilton:
The first is a config option for wccp2 to make squid wait until all
cache_dirs have finished rebuilding before squid will register itself with
WCCP. This will allow the rebuild to happen quickly, and avoid slow web
requests while the
sön 2006-07-30 klockan 09:56 +0200 skrev Guido Serassio:
2006/07/26 16:52:21| ipcCreate: /usr/lib/squid/diskd_daemon: (2) No
such file or directory
A quite large but nonintrusive patch applied to detect this early,
refusing to start at all.
The problem seems to be into ipcCreate():
Yes and
mån 2006-07-03 klockan 21:33 -0300 skrev Giancarlo Razzolini:
Here it is. Please send any comments, critics or suggestions. Did the
patch against the getpwnam.c from 2.6.STABLE1. Also i've started today
writing the documentation for it.
Applied to Squid-3.
Regards
Henrik
signature.asc
lör 2006-07-29 klockan 09:35 +0200 skrev Guido Serassio:
Hi,
I have found an occasional DNS resolution error when browsing
www.microsoft.com.
I have seen the error only few times, less then 10, but the odd thing
is that this happens always only with www.microsoft.com and sometime
it
lör 2006-07-29 klockan 12:47 +0200 skrev Henrik Nordstrom:
lör 2006-07-29 klockan 09:35 +0200 skrev Guido Serassio:
Hi,
I have found an occasional DNS resolution error when browsing
www.microsoft.com.
I have seen the error only few times, less then 10, but the odd thing
lör 2006-07-29 klockan 18:05 +0800 skrev Steven:
I could reproduce the bug if I had a COSS cache_dir enabled without any
aufs cache_dirs. I've updated the bug with a patch to fix this scenario.
I think the COSS issue is separate. Based on your patch that problem
should be seen immediately on
lör 2006-07-29 klockan 09:44 +0200 skrev Jan Engelhardt:
Hello,
Regular client-side transparent proxying is easily accomplished by
redirecting network traffic using -j DNAT, -j REDIRECT, or -j TPROXY (I do
not know why this seems needed). However, server-side transparency requires
a
lör 2006-07-29 klockan 17:05 +0200 skrev Jan Engelhardt:
The relevant parts of the code to fix this is in FwdState::pconnPush and
FwdState::connectStart fwdPconnPool-pop().
What would I have to add?
You would need to extend the key used in these functions with at least
the source IP of the
lör 2006-07-29 klockan 23:16 +0800 skrev Steven:
I was seeing the msgrecv() calls while running strace, but it wasn't in
the same loop as reported in the bug. Looks like I just found another bug
while trying to reproduce this one :)
Was not aware there was msgrcv() calls in pthreads.
We
fre 2006-07-28 klockan 10:55 +0200 skrev Eduard Veleba:
Dear Squid developers,
in our company Seznam.cz we use your application Squid for a long time,
for example at our service http://wiki.mapy.cz,
and we are very satisfied with it. There was recently a demand to cache
pages that has set
fre 2006-07-28 klockan 22:33 +0200 skrev Henrik Nordstrom:
and we are very satisfied with it. There was recently a demand to cache
pages that has set expiration time shorter than
1 minute, that your system doesn´t allow. So we modified the code a
Thanks, but we have already added
Looked at it breafly, but ran out of ideas.
http://www.squid-cache.org/bugs/show_bug.cgi?id=1703
Regards
Henrik
signature.asc
Description: Detta är en digitalt signerad meddelandedel
fre 2006-07-28 klockan 22:24 +0100 skrev Michael Pye:
However this backport doesn't allow you to specifiy the minimum refresh
time in seconds. You may have a situation where you want the minimum
refresh time to be 20 seconds, but still have some other
refresh_patterns that use 40 seconds.
tor 2006-07-27 klockan 11:02 +0200 skrev Henrik Nordstrom:
tor 2006-07-27 klockan 16:40 +0800 skrev Adrian Chadd:
On Thu, Jul 27, 2006, Steven wrote:
They make me use Outlook at work :)
Here I go again.
Cool! I'm happy for you to commit the store_swapmeta patch.
You
Bug #1696 could use a couple of other eyes.
Turns out the router capability processing has never been tested (return
instead of break in a switch statement.. see the bug), and something
seems fishy there (squid complaining on what to me looks like correct
packets)..
See also
tis 2006-07-25 klockan 23:09 +0200 skrev Henrik Nordstrom:
Turns out the router capability processing has never been tested
See also
http://www.squid-cache.org/mail-archive/squid-users/200607/0450.html as
there was some questionmarks while decoding the WCCP2_HERE_I_AM sent by
Squid.
May
ons 2006-07-26 klockan 11:27 +1200 skrev Reuben Farrelly:
Regarding the mask based assignments, wonder how complex mask arrays a
typical equipment supports. In the specs is an unbounded array.
Does this help?
tis 2006-07-11 klockan 14:34 +0200 skrev Luigi Gangitano:
I've just packaged squid-2.6.STABLE1 and squid-3.0.PRE4 for debian,
enabling epoll() support at build time.
This obviously makes squid fail with kernels older than 2.6.x which
are still supported by debian.
During discussions
ons 2006-07-19 klockan 21:53 +0200 skrev Guido Serassio:
I think that Bug #1681 should be also investigated before STABLE2.
The problem is happening on different platforms with the same symptom
with too much frequency.
I will try to do more investigation during the next weekend.
ons 2006-07-19 klockan 15:13 +0200 skrev Martin Stransky:
Hello guys,
As you probably know, we have one RH specific patch to squid.
Yes, found this out by accident some time ago..
It's for
dynamic configuration of filedescriptors number, via. squid.conf. I
reviewed it for the 2.6
The 2.6.STABLE2 release is approaching with many critical bug fixes.
but Bug #1677 kind of a blocker for the release.. not due to it being a
significant Squid bug, but due to it causing some broken web sites to
fail which worked in 2.5..
Regards
Henrik
signature.asc
Description: Detta är en
mån 2006-07-17 klockan 14:43 +0800 skrev Steven Wilton:
I'm compiling under Debian (stable and unstable), and we are still seeing
the fd set limited to 1024. It's coming in the following path:
/usr/include/sys/capability.h
/usr/include/linux/types.h
/usr/include/linux/posix_types.h
Is
mån 2006-07-17 klockan 16:31 +0800 skrev Steven Wilton:
That would be a bit messy. The problem is that those two linux headers
isn't supposed to be included at all in userspace applications (only
kernel). glibc provides it's own types.
I suppose we could use the same glue as Fedora
mån 2006-07-17 klockan 11:04 +0200 skrev Henrik Nordstrom:
You mean something similar to the attached patch?
Yes, plus the above #define and typedef trickery before including
sys/capability.h to shield us from the linux kernel types..
#define _LINUX_TYPES_H
#define _LINUX_FS_H
typedef
sön 2006-07-16 klockan 21:49 +0800 skrev Adrian Chadd:
I've written some code to push logfile writing into an external process,
freeing up the main squid process from the potentially blocking
stdio writes.
Good.
This code isn't ready to be merged into squid-2.6. Its meant more as a
mån 2006-07-17 klockan 09:25 +1000 skrev Robert Collins:
I did a patch for squid 2.x/early 3.0 that did something similar - and
it suffered from dropped log contents during high load... which is
something I never got *satisfactorily* addressed - its the key thing I
think needs considering.
mån 2006-07-17 klockan 09:07 +0800 skrev Adrian Chadd:
Robert: I do remember your work having that annoying problem of dropping
logfile entries. I've got an idea on how to test it (I'll write a script
to throw lots of consecutively-numbered requests at my code and make sure
they all appear in
mån 2006-07-17 klockan 09:07 +0800 skrev Adrian Chadd:
You flush any pending buffers first, right?
This patch doesn't do that yet. It wasn't that important as buffered data
doesn't get lost during a rotate.
The main reason to flush on rotate is to ensure the logs are whole. It
is not
fre 2006-07-14 klockan 16:59 +0200 skrev Guido Serassio:
Hi,
A question regarding the -rdynamic gcc option:
It should go into LDFLAGS, correct ?
Most likely yes.
Regards
Henrik
signature.asc
Description: Detta är en digitalt signerad meddelandedel
ons 2006-07-12 klockan 09:36 -0300 skrev Gonzalo Arana:
Cool, this fixes what some people have been telling me about
FD limits under Squid-2.6 being stuck at 1024. Good find!
If the limit comes from select(2) limits,
It doesn't. It comes from Linux include header stupidities. The glibc
mån 2006-07-03 klockan 20:07 -0300 skrev Giancarlo Razzolini:
First sorry for the late, it were 2 weeks of tests in my university, so
i was busiest that never. Now that i'm on vacation from university, i do
have more free time.
No problem.
Now, to the patch. For doing what you want, i'll
tor 2006-06-29 klockan 10:40 -0300 skrev Gonzalo Arana:
...
} else if (flag == COMM_OK len == 0 !flags.headers_parsed) {
fwd-fail(errorCon(ERR_ZERO_SIZE_OBJECT, HTTP_BAD_GATEWAY));
[..]
} else if (flag == COMM_OK len == 0) {
[..]
if (!flags.headers_parsed)
/*
ons 2006-06-28 klockan 08:32 +0200 skrev Guido Serassio:
No more errors, so it seems that the problem is fixed.
But spun off into more issues... (Bug #1638)..
this connection pinning business has been a bit of a nightmare to get
correct in all configurations.. hopefully no additional gremlins
ons 2006-06-28 klockan 10:19 +0800 skrev Adrian Chadd:
I do keep getting these during polygraph testing:
041.64| Xaction.cc:74: error: 8/10 (c14) premature end of msg body
1151460879.419466# obj:
http://host:8080/w0e4091fe.60215fc4:0006/t01/_000186bd flags:
basic,GET,chb, size:
tis 2006-06-27 klockan 08:16 +0800 skrev Steven Wilton:
In the original code for 2.5, I had the following logic in commResumeFD() to
handle this situation
if(!(F-read_handler) || !(F-epoll_backoff)) {
debug(5, 2) (commResumeFD: fd=%d ignoring read_handler=%p,
tis 2006-06-27 klockan 08:22 +0800 skrev Adrian Chadd:
Which reminds me, I should really spend some time making sensible defaults
when COSS is using AUFS - right now the aiops code doesn't create any threads
because n_aufs_dirs (or whatever it is) is set to 0; one has to override
it by the
tis 2006-06-27 klockan 10:25 +0200 skrev Guido Serassio:
Could you set a breakpoint on the ERR_ZERO_SIZE_OBJECT fwdFail call in
http.c httpReadReply, and print read_sz from there.. (non-optimized
build please..)
Obviously during a failed request ?
Yes.. this code is only reached on such
tis 2006-06-27 klockan 12:39 +0200 skrev Guido Serassio:
Hi,
At 10.40 27/06/2006, Henrik Nordstrom wrote:
tis 2006-06-27 klockan 10:25 +0200 skrev Guido Serassio:
Could you set a breakpoint on the ERR_ZERO_SIZE_OBJECT fwdFail call in
http.c httpReadReply, and print read_sz from
tis 2006-06-27 klockan 13:30 +0200 skrev Henrik Nordstrom:
One question: Are you using NTLM/Negotiate proxy authentication?
If you are then the problem should be fixed now.
Regards
Henrik
signature.asc
Description: Detta är en digitalt signerad meddelandedel
lör 2006-06-24 klockan 08:48 +0800 skrev Adrian Chadd:
I'm seeing it in my polygraph testing under 2.6 + COSS. It may
be related.
Most likely not... there was a nasty bug related to persistent
connections and NTLM authentication..
Regards
Henrik
signature.asc
Description: Detta är en
None critical / blocker for the release, but still
761: diskd unstable
1584: WCCPv2 unable to register with more than one router on Linux (IP
ID IOS issue)
1602: Need to implement TCP fallback for DNS
1638: Connection pinning a bit too agressive. Should not pin due to
local authentication,
mån 2006-06-26 klockan 11:40 +0800 skrev Steven Wilton:
When I did the original serverfd work for backed off connections, I assumed
serverfd=0 was invalid. This patch fixes the code so serverfd=-1 is the
default when there is no backed-off server fd.
Thanks. Applied.
I've also added an
sön 2006-06-25 klockan 17:43 +1200 skrev Doug Dixon:
- Parts of the SSL cleanup. Main requirements being the pending bits
in the comm code, or refactoring solving this I/O operation
independence
differently.
Bug #1633.
- Reasonable read defer management.
Bug #1634
- COSS
sön 2006-06-25 klockan 09:56 +0200 skrev Guido Serassio:
Hi Henrik,
At 23.39 24/06/2006, Henrik Nordstrom wrote:
Could you try collecting a tcpdump -s 1600 of that traffic?
I have done an ethereal capture saved in tcpdump format.
Thanks, but didn't make me very much wiser. But maybe
1101 - 1200 of 2137 matches
Mail list logo