2008/6/28 Henrik Nordstrom <[EMAIL PROTECTED]>:
> Found that one as well. Made very visible by the first, but can be
> temporarily triggered by UDP as well holding up retransmissions for
> about a minute...
>
> Patch in bugzilla, and applied to squid-2. Needs to be forward-ported to
> squid-3.
Ni
On fre, 2008-06-27 at 15:51 +0200, Henrik Nordstrom wrote:
> But not sure about the user visible bug where the DNS queue apparently
> stops being processed. In my tests the queue continued getting processed
> like expected.. or at least I think so..
Found that one as well. Made very visible by th
On 27/06/2008, Henrik Nordstrom <[EMAIL PROTECTED]> wrote:
> It at least triggers one bug, the high rate and never ending
> retransmission.
>
> But not sure about the user visible bug where the DNS queue apparently
> stops being processed. In my tests the queue continued getting processed
> lik
On fre, 2008-06-27 at 13:41 +0100, Bradley Kite wrote:
> I have another failure, this time with the extra DNS debug output.
>
> I have just reliased that our load balancer, which also load-balances
> DNS lookups, was not properly configured to support TCP (only UDP DNS
> lookups). I think this mi
On 26/06/2008, Henrik Nordstrom <[EMAIL PROTECTED]> wrote:
> On tor, 2008-06-26 at 11:08 +0100, Bradley Kite wrote:
> > I am just rolling out your patch, I will send over the mgr:idns output
> > when I have the next failure. In the mean time here is the other
> > output you requested. The number
On tor, 2008-06-26 at 11:08 +0100, Bradley Kite wrote:
> I am just rolling out your patch, I will send over the mgr:idns output
> when I have the next failure. In the mean time here is the other
> output you requested. The number of callbacks is just 1 - I would
> expect it to be equal to the numbe
On 26/06/2008, Henrik Nordstrom <[EMAIL PROTECTED]> wrote:
> The attached patch makes mgr:idns show a bit more detail about the
> queue.
>
> And one more thing to check in the cachemgr data.
>
> squidclient mgr:events | grep idns
>
> do that give you a idnsCheckQueue entry? (it should, when th
The attached patch makes mgr:idns show a bit more detail about the
queue.
And one more thing to check in the cachemgr data.
squidclient mgr:events | grep idns
do that give you a idnsCheckQueue entry? (it should, when there is
entries in the queue...)
Note: the test above does not require the
On ons, 2008-06-25 at 21:53 +0100, Bradley Kite wrote:
> Internal DNS Statistics:
>
> The Queue:
>DELAY SINCE
> ID SIZE SENDS FIRST SEND LAST SEND
> -- - -- -
> 0x2dd7 39 2597011 40711.226 0.002
> 0x10ec 40 2597905 42519.743 0.
On 25/06/2008, Henrik Nordstrom <[EMAIL PROTECTED]> wrote:
> On ons, 2008-06-25 at 22:28 +0200, Henrik Nordstrom wrote:
>
> > Could be, and that missing cachemgr page I mentioned would show if this
> > is the case..
>
>
> Actually it's already implemented. Silly me.
>
> squidclient mgr:idns
>
>
On ons, 2008-06-25 at 22:28 +0200, Henrik Nordstrom wrote:
> Could be, and that missing cachemgr page I mentioned would show if this
> is the case..
Actually it's already implemented. Silly me.
squidclient mgr:idns
Regards
Henrik
signature.asc
Description: This is a digitally signed message p
On ons, 2008-06-25 at 21:10 +0100, Bradley Kite wrote:
> This server had traffic disabled for at least 30 minutes, so any
> "pending" DNS lookups should have definitely failed/timed-out by the
> time I got the debug.
Ok, THAT is an important hint.
> Maybe its possible that somehow these DNS look
On 25/06/2008, Henrik Nordstrom <[EMAIL PROTECTED]> wrote:
> On ons, 2008-06-25 at 12:30 +0100, Bradley Kite wrote:
> > I dont know the code that well but if the entry is not in the ipcache,
> > then how can idnsCachedLookup return true?
>
>
> This happens if there is already a pending matching p
On ons, 2008-06-25 at 12:30 +0100, Bradley Kite wrote:
> I dont know the code that well but if the entry is not in the ipcache,
> then how can idnsCachedLookup return true?
This happens if there is already a pending matching pending DNS lookup.
The idns cache is a cache of pending DNS lookups, not
On 24/06/2008, Henrik Nordstrom <[EMAIL PROTECTED]> wrote:
> tis 2008-06-24 klockan 22:50 +0800 skrev Adrian Chadd:
>
> > The question is - why is Squid not timing the client-facing filedescriptors
> > out? Is there some pending reply which isn't being written back correctly?
>
>
> Good question.
tis 2008-06-24 klockan 22:50 +0800 skrev Adrian Chadd:
> The question is - why is Squid not timing the client-facing filedescriptors
> out? Is there some pending reply which isn't being written back correctly?
Good question.
To answer that we need to look into in detail what is going on with the
On 23/06/2008, Adrian Chadd <[EMAIL PROTECTED]> wrote:
> On Mon, Jun 23, 2008, Bradley Kite wrote:
> > Hi all,
> >
> > I have a problem that I am currently trying to diagnose. Please
> > correct me if you think this should be on the -users list instead, but
> > I think its a bug and I need fur
On Tue, Jun 24, 2008, Bradley Kite wrote:
> I have some debug for you now:
Argh, just before my exams! God damned users you.
> This server just started showing the symptoms described earlier. When
> dropped into debug mode, the defer count is mostly around 6 or 7
> thousand - each being 1 second
On 23/06/2008, Henrik Nordstrom <[EMAIL PROTECTED]> wrote:
> On mån, 2008-06-23 at 12:18 +0100, Bradley Kite wrote:
>
> > I am concerned that, for which ever reason, squid stops processing
> > requests for a particular website, and then fails to detect when
> > clients give up, incorrectly putti
On mån, 2008-06-23 at 12:18 +0100, Bradley Kite wrote:
> I am concerned that, for which ever reason, squid stops processing
> requests for a particular website, and then fails to detect when
> clients give up, incorrectly putting the FD into the "half-closed"
> state, leading to the situation wher
On Mon, Jun 23, 2008, Bradley Kite wrote:
> Hi all,
>
> I have a problem that I am currently trying to diagnose. Please
> correct me if you think this should be on the -users list instead, but
> I think its a bug and I need further help trying to diagnose it.
Ruh roh!
> Dropping the squid server
Hi all,
I have a problem that I am currently trying to diagnose. Please
correct me if you think this should be on the -users list instead, but
I think its a bug and I need further help trying to diagnose it.
We run an array of 30 squid servers, sitting behind a load balancer.
>From time-to-time t
22 matches
Mail list logo