I tried to update the 1.2.5 Windows source I have with all the changes that went into 1.2.6, but it didn't compile and I'm completely unfamiliar with C programming on Windows so I have no idea why it didn't work, and not the skill to make it work. It would be awesome with an active windows maintainer and to get all the Windows changes into the main version, but I'm clearly not that person. Sorry. :-)
/Henrik Schröder On Sun, Aug 24, 2008 at 8:45 PM, dormando <[EMAIL PROTECTED]> wrote: > > Can you look at the change history from 1.2.5 -> 1.2.6 and apply the CAS > crash fix? Maybe the rest of the crash fixes too? I don't know how hard > it'll be to update the windows port to 1.2.6, but someone really ought > to... > > -Dormando > > On Sun, 24 Aug 2008, Henrik Schr?der wrote: > > > Yes, I did the changes in our client, and it works fine again now. > > > > I'll see if I can make a test for memcached that consistently triggers > this > > error, it seems like it happens when you do a "gets" on a lot of keys > > (100+). That ought to work just as well as doing a "get" on those same > keys. > > > > > > /Henrik > > > > On Sun, Aug 24, 2008 at 12:53 PM, Simone Busoli <[EMAIL PROTECTED] > >wrote: > > > >> Fixed? > >> > >> > >> On Tue, Aug 19, 2008 at 1:25 PM, Henrik Schr?der <[EMAIL PROTECTED] > >wrote: > >> > >>> Yeah, uh, I'm running version 1.2.5 on Windows, that's probably useful > to > >>> know, I should not have forgotten that in my original email. :-) > >>> > >>> I'm thinking that we get the error because of a recent change in our > >>> client where we do "gets" everywhere, and we do have some massive > >>> multi-gets, and retrieveing hundreds of values along with their CAS > unique > >>> in a single get might cause some resource starvation. > >>> > >>> I'm fixing our client now to properly use "gets" only when needed, and > >>> we'll see if that helps. > >>> > >>> > >>> /Henrik > >>> > >>> > >>> On Tue, Aug 19, 2008 at 11:52 AM, Henrik Schr?der <[EMAIL PROTECTED] > >wrote: > >>> > >>>> Hi everyone, > >>>> > >>>> A short while ago, one of our memcached servers started dying, and if > I > >>>> run it in verbose mode (-vv) it prints "Failed to write, and not due > to > >>>> blocking" before exiting. I can find the place in the source where it > does > >>>> this, but the error message tells me very little, and I'm not familiar > with > >>>> the memcached code, could someone else fill me in on what exactly goes > >>>> wrong? Some help on how to avoid the error would be nice as well... > :-) > >>>> > >>>> > >>>> /Henrik > >>>> > >>> > >>> > >>> > >>> > >> > >>> > >> > > > > > > > > > > >
