Client crash hangs entire backup job?
*This message was transferred with a trial version of CommuniGate(r) Pro* Server 2.4.5, running on Gentoo Clients mix of 2.4.4 and 2.4.5, amanda client on various Linux flavors and Windows using the amanda client via Cygwin One of my Windows servers is having an intermittent problem and has been crashing (sometimes) when it's trying to start up with backups. It's a W2K server, Cygwin, doing a fairly selective backup of some directories using svr-comp-user-tar. From what I can see in the logs, it's creating the estimate and sending it off to the server. It looks like when it actually starts doing the backup the machine completely dies. I'm thinking I'm going to remove Cygwin and redo the whole amanda install on it, but my question is really about the server end of things. The server continues to show it's getting the estimate from the client, and never gives up and moves on to doing the rest of the machines. Wouldn't be that big a deal, except I've got 35 servers being backed up, and one failure bringing down the whole lot is a bit of an issue. Machines that just don't answer at all to the initial request for an estimate don't stop things, amanda just notes she didn't get a response and moves on. Are there any configuration tweaks I can do to make it go forward instead of just halting everything? Thanks, LP
Re: Amanda on Gentoo Linux?
My backup server is Gentoo, as are a couple of other servers and workstations it backs up. It's been running steady as a rock for years. The only issue we've hit recently is not being able to get incrementals on machines with updated tar packages. We get Level 0 every night. We stay pretty on top of patches from Gentoo, and we've seen it on multiple systems. We're also seeing it on Windows 2003 boxes running amanda under Cygwin. I don't think that has anything to do with the server, only with the client machine(s). We are still at 2.4.5. LP
Re: Oracle and mySql backups?
For our mySql and mssql backups we dump the databases an hour or so prior to the start of the backup cycle. For mySql it's a bash script launched by a cron job, for mssql it's a backup job scheduled through EM. We generally keep about a weeks worth on the disk. Our amanda server only backs up the sql dump directory. It's quicker and easier for us to install a server and restore the data than to do a bare bones recover. LP
Re: FAILURE AND STRANGE DUMP SUMMARY
We use the following script (via a cron job) to get a distinct file for each database. We then only back up the /var/spool/mysqldumps directory. We run the script about an hour before the backups kick off. LP #!/bin/bash # Daily database dumps for inclusion in amanda backups. MYU="/mysql user with appropriate permissions/" MYP="/mysql users password/" MYDBS=`mysql -e "show databases;" -u $MYU -p$MYP|sed -e 's/|//g' |sed -e '1d'` for d in $MYDBS; do mysqldump --user=$MYU --password=$MYP $d > /var/spool/mysqldumps/$d.sql done
Re: FAILURE AND STRANGE DUMP SUMMARY
fedora wrote: > Almost files which is the extension ./databasename/*.MYI changed not only > ./mysql/general_log.CSV > Of course they changed. I usually try to avoid "Read the documentation" type responses, but I think this might be a good time to suggest it. The mysql docs tell you pretty clearly that backing up the database files directly is not going to work. You need to either dump the databases and back those up (a number of methods have been presented) or use a tool specifically designed to do them on the fly. Dustin posted a link to that a couple of days ago. LP
View Compile Options?
I'm finally going to upgrade our amanda installation from 2.4.5 to 2.5.2. It's been running steady as a rock for years, but I've reached a point where I either have to replace drives/tapes, or I have to be able to span tapes with a DLE. With money tight - the answer is to upgrade. Because the server has run solid for so long - since before I came here - I have no idea what options it was compiled with. There are no notes anywhere. It's long since scrolled out of .bash_history. Is there a way to see which options were compiled in? Thanks, LP
Re: View Compile Options?
Jean-Francois Malouin wrote: > * Linda Pahdoco <[EMAIL PROTECTED]> [20070906 15:46]: > >> I'm finally going to upgrade our amanda installation from 2.4.5 to >> 2.5.2. It's been running steady as a rock for years, but I've reached a >> point where I either have to replace drives/tapes, or I have to be able >> to span tapes with a DLE. With money tight - the answer is to upgrade. >> >> Because the server has run solid for so long - since before I came here >> - I have no idea what options it was compiled with. There are no notes >> anywhere. It's long since scrolled out of .bash_history. >> >> Is there a way to see which options were compiled in? >> > > Look for the amandad debug files. > jf > > >> Thanks, >> LP >> > > That pointed me in the right direction, thank you. I actually found them in the amdump log. LP
Tape Spanning
Dustin J. Mitchell wrote: > > That arrangement is ok, with the proviso that you can't use tape > spanning on the server, or you won't be able to recover using older > clients. There may be more limitations -- someone else please chime > in? > > Dustin > > Changing the subject so as not to hijack the thread So just upgrading the server to 2.5.x won't buy me tape spanning? The server and client both have to be done? Not a huge deal since I've just got a couple of hosts that are to the point of needing the capability, but I just want to be sure I understand what I'm getting into. Thanks, LP
Restore from file, not tape
I was really hoping I could get the tape changer configured before this happened, but it bit me on the rear. Amanda 2.4.5 server. I've got a level 0 sitting on my disk. It's too large to fit on tape. I need to extract a directory from it. I know the entire path. It doesn't show up in my list on amrestore because it never got dumped to tape from the holding disk. This has probably been covered before and I just didn't search right - but does someone know the commands to pull the directory from the file? Thanks, LP
Re: Restore from file, not tape
Jon LaBadie wrote: > Amrecover, not amrestore, would be the command to extract a directory > tree from that holding disk (or tape). > > If that is what you used, and you were unable to see the index of files > in the dump, was the "record" parameter set to "yes" when the dump was > made. No index is created otherwise. If it was set, show the command > and interactive amrecover session you used (see "script" command as a > possible recorder) > Sorry, my mistype. I am using amrecover. When I dump the history for the DLE I get this: amrecover> history 200- Dump history for config "DailySet1" host "" disk "" 201- 2007-10-30 3 DailySet1-051 72 201- 2007-10-29 3 DailySet1-047 75 201- 2007-10-28 2 DailySet1-042 77 201- 2007-10-27 2 DailySet1-037 67 201- 2007-10-26 1 DailySet1-033 4 201- 2007-10-24 1 DailySet1-027 67 201- 2007-10-22 4 DailySet1-022 5 201- 2007-10-21 4 DailySet1-015 4 201- 2007-10-20 4 DailySet1-010 4 201- 2007-10-19 4 DailySet1-006 4 201- 2007-10-18 4 DailySet1-117 5 201- 2007-10-16 3 DailySet1-112 4 201- 2007-10-14 3 DailySet1-103 76 201- 2007-10-13 2 DailySet1-099 68 201- 2007-10-12 2 DailySet1-094 74 201- 2007-10-11 1 DailySet1-093 4 201- 2007-10-10 1 DailySet1-083 59 201- 2007-10-07 5 DailySet1-072 68 My level 0 ran on the 23rd, and is sitting on my disk. As you can see, it doesn't show in the dump history, I'm guessing because it didn't actually get dumped onto a tape. I can (and did) use "settape" to point to the file. However, since the dump isn't showing in my history I can't tell amanda to use it. If I choose one of the other dates I do see the correct file list. I'm going to go read on dd/tar, but if I had a way to use amrecover that's what I'd like to do. Thanks, LP
Re: Restore from file, not tape
I'm manually extracting the files now based on the clues you all gave me. I'll come back to what broke once I get the customer back up. Thank you all for your suggestions. As usual - you're very helpful. :) LP
