4.10 crash

classic Classic list List threaded Threaded
4 messages Options
Reply | Threaded
Open this post in threaded view
|

4.10 crash

Dmitry Melekhov
Hello!

see this in log, after this squid dies.

2020/02/26 11:17:47 kid1| UPGRADE WARNING: URL rewriter reponded with
garbage ' 192.168.23.54/ABDRASHITOV-RR.p98a3.belkam.com abdrashitovrr
CONNECT myip=192.168.22.254 myport=8090'. Future Squid wil
l treat this as part of the URL.
2020/02/26 11:17:48 kid1| assertion failed: MemBuf.cc:354: "new_cap >
(size_t) capacity"


from systemctl status:


фев 26 11:17:48 inetgw2 squid[1167]: Squid Parent: squid-1 process 11576
exited due to signal 6 with
фев 26 11:17:48 inetgw2 squid[1167]: Squid Parent: squid-1 process 11576
will not be restarted for 3
фев 26 11:17:48 inetgw2 squid[1167]: Exiting due to repeated, frequent
failures

I see this several times, have no idea what  caused this, even if this
is redirector error, squid should not crash...

Is there any way to fix this?


Thank you!



_______________________________________________
squid-users mailing list
[hidden email]
http://lists.squid-cache.org/listinfo/squid-users
Reply | Threaded
Open this post in threaded view
|

Re: 4.10 crash

Dmitry Melekhov
26.02.2020 11:42, Dmitry Melekhov пишет:

> Hello!
>
> see this in log, after this squid dies.
>
> 2020/02/26 11:17:47 kid1| UPGRADE WARNING: URL rewriter reponded with
> garbage ' 192.168.23.54/ABDRASHITOV-RR.p98a3.belkam.com abdrashitovrr
> CONNECT myip=192.168.22.254 myport=8090'. Future Squid wil
> l treat this as part of the URL.
> 2020/02/26 11:17:48 kid1| assertion failed: MemBuf.cc:354: "new_cap >
> (size_t) capacity"
>
>
> from systemctl status:
>
>
> фев 26 11:17:48 inetgw2 squid[1167]: Squid Parent: squid-1 process
> 11576 exited due to signal 6 with
> фев 26 11:17:48 inetgw2 squid[1167]: Squid Parent: squid-1 process
> 11576 will not be restarted for 3
> фев 26 11:17:48 inetgw2 squid[1167]: Exiting due to repeated, frequent
> failures
>
> I see this several times, have no idea what  caused this, even if this
> is redirector error, squid should not crash...
>
> Is there any way to fix this?
>
>
> Thank you!
>
>
>
btw, looks like here is request, which causes problem - time can be
different, we hit this several times...


1582701916.487     34 192.168.23.54 TCP_MISS/404 723 GET
http://detectportal.firefox.com/success.txt192.168.23.54/ABDRASHITOV-RR.p98a3.belkam.comabdrashitovrrGETmyip=192.168.22.254myport=8090 
abdrashitovrr HIER_DIRECT/212.188.32.137 application/xml

_______________________________________________
squid-users mailing list
[hidden email]
http://lists.squid-cache.org/listinfo/squid-users
Reply | Threaded
Open this post in threaded view
|

Re: 4.10 crash

Amos Jeffries
Administrator


On 26/02/20 8:46 pm, dm wrote:
> 26.02.2020 11:42, Dmitry Melekhov пишет:
>> Hello!
>>
>> see this in log, after this squid dies.
>>
>> 2020/02/26 11:17:47 kid1| UPGRADE WARNING: URL rewriter reponded with
>> garbage ' 192.168.23.54/ABDRASHITOV-RR.p98a3.belkam.com abdrashitovrr
>> CONNECT myip=192.168.22.254 myport=8090'. Future Squid wil
>> l treat this as part of the URL.

And guess what?

From your access.log this is the URL:

>
http://detectportal.firefox.com/success.txt192.168.23.54/ABDRASHITOV-RR.p98a3.belkam.comabdrashitovrrGETmyip=192.168.22.254myport=8090


... so the helper is one thing you definitely need to get fixed.


>> 2020/02/26 11:17:48 kid1| assertion failed: MemBuf.cc:354: "new_cap >
>> (size_t) capacity"


This happened up to two seconds after the previous log entry. A great
many transactions may have passed in between the two occurances.


>>
>>
>> from systemctl status:
>>
>>
>> фев 26 11:17:48 inetgw2 squid[1167]: Squid Parent: squid-1 process
>> 11576 exited due to signal 6 with
>> фев 26 11:17:48 inetgw2 squid[1167]: Squid Parent: squid-1 process
>> 11576 will not be restarted for 3
>> фев 26 11:17:48 inetgw2 squid[1167]: Exiting due to repeated, frequent
>> failures
>>
>> I see this several times, have no idea what  caused this, even if this
>> is redirector error, squid should not crash...
>>
>> Is there any way to fix this?
>>

All the info we have now on the assertion is that the capacity value for
a memory buffer is trying to be grown into a negative size range (32-bit
overflow?). We will need a stack trace from assertion/crash to know
which buffer is being expanded and maybe find out why.

Details on how to get a stack traces are documented at
<https://wiki.squid-cache.org/SquidFaq/BugReporting>. That includes from
core files, or from a running proxy without affecting uptime any worse
than the crash already does.


Amos
_______________________________________________
squid-users mailing list
[hidden email]
http://lists.squid-cache.org/listinfo/squid-users
Reply | Threaded
Open this post in threaded view
|

Re: 4.10 crash

Alex Rousskov
In reply to this post by Dmitry Melekhov
On 2/26/20 2:42 AM, Dmitry Melekhov wrote:
> 2020/02/26 11:17:48 kid1| assertion failed: MemBuf.cc:354: "new_cap >
> (size_t) capacity"

This is most likely a Squid bug. I recommend filing a bug report with
Squid Bugzilla and providing the details that Amos has mentioned in his
response.


Thank you,

Alex.
_______________________________________________
squid-users mailing list
[hidden email]
http://lists.squid-cache.org/listinfo/squid-users