Hi Willy, Thanks for you detail information, I set the core dump file to unlimited and try to get it if it hang again. and then I will try the 32bit and give your feedback if it hang again later.
Best Regards, Hogan On Fri, Nov 19, 2010 at 4:08 AM, Willy Tarreau <[email protected]> wrote: > Hi Hogan, > > On Thu, Nov 18, 2010 at 04:18:45PM +0800, Hogan Yu wrote: > > Hi Willy, > > > > I catch a new big problem about the Haproxy after it runs two days > > successfully > > I add a command in my config file > > > > reqirep ^([^\ ]*)\ /loot/(.*) \1\ /bigloot/tutorial/index/\2 > > > > it replace the path /loot/att to /bigloot/tutorial/index/att. > > > > Then the Haproxy hangs every hours and I get the error as follows: > > Nov 17 23:11:31 239488-web3 kernel: haproxy[11343]: segfault at > > 00000000203ab000 rip 0000003ef9c7c51a rsp 00007fff7fa050a8 error 6 > > Huh ? This is a critical error. It should absolutely never happen whatever > the configuration, it is necessarily a bug. > > > I google and can not find any answer for that. > > because it normally never happens :-/ > > > Here is some information about my haproxy. > > Thanks, very much appreciated. > (...) > > Nothing looks strange in your config. > > > do you know what I can do for that? does the reqirep problem? > > I don't know. The few things that come to mind that are not very > common in other configs : > > - reqirep matching a string ending in .* > - cookie+appsession+rewrite > > But that should not be a reason for a crash. > > If it is still running, what would immensely help would be to run > it by doing this before starting it : > > # ulimit -c unlimited > > It will dump a core which will tell us where it crashes. DO NOT POST > the core to the list, it may contain sensitive information such as > the requests in flight at the instant of the crash. However, it will > be of huge value to try to understand if a memory area was overwritten > or something like this. Also, please provide your haproxy binary along > with it, since the core is only exploitable with the binary. > > You can also try to rebuild it in 32-bit mode, just in case the issue > only happens in 64-bit. We never know, that was once the case in the > past. > > Going back to another version could help you get your site running > again, but given that we did not change anything in the rewrite code > recently, I have little hopes that it could change anything. > > Best regards, > Willy > >
