New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
gbak intermittantly reports "Fatal lock manager error: invalid lock id (10963360), errno: 22" in firebird.log and the backup fails [CORE4022] #4352
Comments
Commented by: @aafemt That's result of running gbak in embedded mode from root. Don't ever do it. And discover firebird-support mail list to yourself. |
Commented by: Fox (fox7x) Thanks for responding, will have any recommendations for my backup process correctly? |
Commented by: @dyemanov Your Firebird setup is wrong, fb_lock_mgr must be running as root. It will resolve the issue. |
Commented by: @pmakowski fb_lock_mgr as root ? nope, I don't think so |
Modified by: @pcisarstatus: Resolved [ 5 ] => Closed [ 6 ] |
Commented by: @dyemanov Philippe, this is the only way the lock manager can work with both deamon-spawned and embedded connections at the same time, regardless of the OS user gbak/isql/whatever is running under. Otherwise, signals cannot be delivered from "firebird" owned sessions to "root" owned ones. So other suggestions are just workarounds. And AFAIK, our install scripts setup Firebird exactly this way. |
Commented by: @pmakowski ok and in fact it was the case in previous EPEL packages in the spec there was a : so please Fox report the bug in Red Hat Bugzilla if need |
Commented by: Fox (fox7x) Ok, then I assume that the bug is EPEL RPMs?. If I install firebird on the same OS using "FirebirdCS-2.1.5.18496-0.amd64.tar.gz", the problem should be solved? |
Commented by: @pmakowski yes bug in EPEL rpms, please report it to Red Hat Bugzilla. |
Commented by: Fox (fox7x) OK, thank you. |
Commented by: @pmakowski package update done : |
Commented by: Fox (fox7x) Thank you Philippe. Is not yet in EPEL (using yum check-update). I look to publish it in the EPEL repository? |
Commented by: @pmakowski yes it will take some time, I just built it, then asked it to be push in testing and after few days in testing or karma 3, it will go to update, that the process, but all that it will do is : so you can do it now by your self on your server |
Commented by: Fox (fox7x) Ok Philippe. Thank you very much again. I'll keep you updated as behaves my installation of firebird with this change. I really appreciate your quick response to this situation. |
Commented by: Fox (fox7x) Philipe, I apply to my server the command chmod 6550 on /usr/lib64/bin-classic/fb_lock_mgr, but the process fb_lock_mgr do not see in ps or top, later i modify the file /etc/xinetd.d/firebird , i change the parameter user from firebird to root. Is normal this or my file /etc/xinetd.d/firebird is wrong? |
Commented by: @dyemanov fb_lock_mgr is running (and hence visible) as long as you have active connections to databases. Running firebird itself as root (what you did in the xinetd config) is a bad idea due to potentially compromised security. |
Commented by: Fox (fox7x) The situation that I mention above is with connections to the database. Even with connections to the database I don't see fb_lock_mgr. To fix it I had to modify fb_inet_server to root user (in the file /etc/xinetd.d/firebird), only in this way appeared the fb_lock_mgr process. |
Commented by: Fox (fox7x) To avoid setting the root user in the file /etc/xinetd.d/firebird and see the fb_lock_mgr process, I add the following command first: chown root:firebird /usr/lib64/bin-classic/fb_lock_mgr and then chmod 6550 /usr/lib64/bin-classic/fb_lock_mgr. This based on how leaves the same file to install firebird using the FirebirdCS-2.1.5.18496-0.amd64.tar.gz. You can help me with this?. Is this correct or am I missing something else?. fb_lock_mgr when the tar.gz from http://firebirdsql.org is used: fb_lock_mgr when only EPEL is used: fb_lock_mgr when only using EPEL and chmod 6550 (here I can not see the fb_lock_mgr process): fb_lock_mgr when using EPEL, chown root:firebird and chmod 6550 (here is when everything works apparently fine and i see the fb_lock_mgr process): Notes: You can support me with this situation please?. Why using EPEL everything belongs to root and using the .tar.gz from http://firebirdsql.org everything belongs to firebird?. |
Commented by: @pmakowski Fox : fb_lock_mgr is now correctly set : and update is coming, but it take some times (distro rules, an update have to wait 2 weeks in testing before being a regular update) |
Commented by: Fox (fox7x) Ok Philipe, thanks, sorry for the inconvenience. It was just that you confirm me if on my server that use the two commands mentioned in the last comment (chown root:firebird and chmod 6550 over fb_lock_mgr) so that everything works correctly, because as you said the update is in EPEL testing. It was only a request for confirmation or recommendation to apply on my server before it reaches the upgrade of EPEL. |
Submitted by: Fox (fox7x)
The bug occurs intermittently when run a backup script (gbak on cron) every day at night.
The script of backup contains the following:
cd /var/lib/firebird/data
rm -f res_MAINDBI.log
gbak -g -b -z -v -user "sysdba" -pas "masterkey" -y res_MAINDBI.log MAINDB1.FDB MAINDBI.FBK
The error does not occur every night, just some, when it occurs, the backup fails,
the back-end of the log does not say the words "gbak:closing file, committing, and finishing. 2665691648 bytes written" so that the backup is not successful.
When this bug occurs in "firebird.log" file shows the following:
srvcentos.local Wed Dec 19 01:00:08 2012
ISC_kill: process 25266 couldn't deliver signal 16 to process 10141: permission denied
srvcentos.local Wed Dec 19 01:00:08 2012
Fatal lock manager error: invalid lock id (10963360), errno: 22
Notes:
OS: Centos 5.8 x86_64 updated to date.
Firebird Classic Server: This is the version 2.1.5 of the EPEL repository.
Installed Packages from EPEL (http://dl.fedoraproject.org/pub/epel/5/x86_64/repoview/firebird-classic.html):
firebird.x86_64 [2.1.5.18496.0-2.el5]
firebird-classic.x86_64 [2.1.5.18496.0-2.el5]
firebird-filesystem.x86_64 [2.1.5.18496.0-2.el5]
firebird-libfbclient.x86_64 [2.1.5.18496.0-2.el5]
firebird-libfbembed.x86_64 [2.1.5.18496.0-2.el5]
Runing fb_inet_server proccess as user "firebird" and the script as user "root" using cron:
Process A
USER: firebird
COMMAND: fb_inet_server
Process B
USER: firebird
COMMAND: fb_lock_mgr
Process C
USER: root
COMMAND: cron (script of backup)
The bug only occurs every few weeks, but when it happens, the backup fails.
The text was updated successfully, but these errors were encountered: