For what it's worth, we use -verify_all to sync the master server at each data center, then we rsync the contents to all of the slaves.
I don't have any theories what might be different about steam.inf, other than failure to update that particular file is so highly visible, whereas failure for other files might go undetected. The game master server (GMS), which does the server listing and also is responsible for delivers the message, "Your server is out of date" is completely distinct from our content servers, which actually deliver the game bits. When we release an update we first send out the bits to the content servers. Then, when all of the content servers have all of the bits, then we flip the master switch to make those depots available for download. After a short delay, this also lets the GMS know that there is a new version of the product. -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of PharaohsPaw Sent: Friday, December 16, 2011 1:05 PM To: Half-Life dedicated Linux server mailing list Subject: Re: [hlds_linux] (ATTN: Valve) steam.inf missing and other "incomplete update" problems Hi Fletch, No, it's OK. Thanks for responding. I do have to wonder if there is something I could, or SHOULD do differently. With folks like Steve who posted a few replies back mentioning that he updates 145 servers and has never seen this problem before - and that he doesn't use -verify_all either, it does make me wonder what could possibly be wrong, or at least "fatally different" about what our setup does. The steam.inf file must update properly for some people, since Steve has never seen a problem (and he has a lot more servers being updated also). Steve how do your servers update? Do you have -autoupdate on the command line? Do they just update when the map changes, or are you doing something different such as one of the plugins that uses steamtools to handle the master request restart? I'd hate to imagine having to manually restart/update 145 servers. :) I'm also wondering if maybe what I should do is edit the nemrun script to put some kind of a delay in, so that when an update is detected (because srcupdatecheck returned a 7 code), it waits a little while before it actually begins the update process (calling steam to update the tree) -- so that even though we detect an update being released really quickly after it hits, and knows it needs to start an update, it waits X number of seconds first - to try to give extra time for the files to be pushed to the masters before we start that update. If the problem has anything to do with waiting long enough for the update files to be propagated to any master the steam binary is likely to hit to begin the update, this would probably help. Fletch is there any chance at all that the steam.inf file might just be one of the last files to get placed onto the masters when update files are loaded/replicated? If so, then this could be another indicator that nemrun just needs to wait a little longer before it starts an autoupdate. Thanks, PharaohsPaw _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: https://list.valvesoftware.com/cgi-bin/mailman/listinfo/hlds_linux _______________________________________________ To unsubscribe, edit your list preferences, or view the list archives, please visit: https://list.valvesoftware.com/cgi-bin/mailman/listinfo/hlds_linux
