cancel
Showing results for 
Search instead for 
Did you mean: 

Catch 22 with new database having a full log

Former Member
0 Kudos

We need some help. One of our new non-production MaxDB databases has run into the situation where its log filled up before an initial backup was taken. Since no backup has been taken, we cannot activate autolog to resolve the full log situation. When we try to take a backup, we receive the following two errors:

-24755,ERR_SESSIONLIMIT: No session of type 'User' available

-24994,ERR_RTE: Runtime environment error 2,task limit

We presume this is because the database is not accepting any more activity due to the full log. How can we get a backup complete so we can then activate auto logging? If we can't get a backup at this point, how can we get the the autolog activated, or even set the log overwrite on temporarily to get us unstuck and allow the backup?

Any suggestions would be extremely welcome.

Thanks,

Clint

Accepted Solutions (0)

Answers (3)

Answers (3)

Former Member
0 Kudos

Closing out question. Thanks to Lars and Natalia for their fine assistance.

Former Member
0 Kudos

Lars, Natalia,

Thank you very much for all your help. We noticed that the output from the "x_cons <db_name> show active" command resulted in 50 tasks being displayed, which was the limit set for MAXUSERTASKS for this database. We were able to increase that number dynamically in Database Manager. This then allowed us to connect in the normal fashion to obtain the initial backup.

I also want to try the utility connection to see if that would have worked as well. We were sort of heading down that road before opening this thread, but were evidently doing something incorrectly.

Once again, thank you both for your help. You have no idea how much it is appreciated as we have been having to learn MaxDB on the fly with no kind of training whatsoever.

lbreddemann
Active Contributor
0 Kudos

> we have been having to learn MaxDB on the fly with no kind of training whatsoever.

Hi Clint,

unfortunately this seems to be a very common thing.

People are expected to get a grasp on thing just on the fly.

Maybe the available training material in the [MaxDB Wiki|https://www.sdn.sap.com/irj/scn/wiki?path=/display/maxdb/main] and on the [MaxDB Website|http://maxdb.sap.com/] provide some more help to you and your colleagues.

best regards and good luck with MaxDB.

Lars

former_member229109
Active Contributor
0 Kudos

Hello Clint,

-> What is the version of your database?

-> Please update with output of the command:

x_cons <DB-Name> show active

Thank you & best regards, Natalia Khlopina

Former Member
0 Kudos

Hi Natalia,

The database version is 7.6.00.18. Following is the output of the x_cons <db_name> show active command:

dwic694:sqdqn2 2> x_cons QN2 show active

SERVERDB: QN2

ID UKT UNIX TASK APPL Current Timeout Region Wait

tid type pid state priority cnt try item

T113 7 -1 User 815332* LOG FULL (246) ! 0 0 32635808(s)

T114 7 -1 User 1245276* LOG FULL (246) ! 0 0 32635808(s)

T115 7 -1 User 868442* LOG FULL (246) ! 0 0 32635808(s)

T116 7 -1 User 1118238* LOG FULL (246) ! 0 0 32635808(s)

T117 7 -1 User 528524* LOG FULL (246) ! 0 0 32635808(s)

T118 7 -1 User 708720* LOG FULL (246) ! 0 0 32635808(s)

T119 7 -1 User 1069188* LOG FULL (246) ! 0 0 32635808(s)

T120 7 -1 User 434266* LOG FULL (246) ! 0 0 32635808(s)

T121 7 -1 User 1126500* LOG FULL (246) ! 0 0 32635808(s)

T122 7 -1 User 921694* LOG FULL (246) ! 0 0 32635808(s)

T123 7 -1 User 717052* LOG FULL (246) ! 0 0 32635808(s)

T124 7 -1 User 594026* LOG FULL (246) ! 0 0 32635808(s)

T125 7 -1 User 831508* LOG FULL (246) ! 0 0 32635808(s)

T126 7 -1 User 667788* LOG FULL (246) ! 0 0 32635808(s)

T127 7 -1 User 393338* LOG FULL (246) ! 0 0 32635808(s)

T128 7 -1 User 860244* LOG FULL (246) ! 0 0 32635808(s)

T129 7 -1 User 823354* LOG FULL (246) ! 0 0 32635808(s)

T130 7 -1 User 905424* LOG FULL (246) ! 0 0 32635808(s)

T131 7 -1 User 889016* LOG FULL (246) ! 0 0 32635808(s)

T132 7 -1 User 733208* LOG FULL (246) ! 0 0 32635808(s)

T133 7 -1 User 1040390* LOG FULL (246) ! 0 0 32635808(s)

T134 7 -1 User 983220* LOG FULL (246) ! 0 0 32635808(s)

T135 7 -1 User 614530* LOG FULL (246) ! 0 0 32635808(s)

T136 7 -1 User 368752* LOG FULL (246) ! 0 0 32635808(s)

T137 7 -1 User 1024044* LOG FULL (246) ! 0 0 32635808(s)

T138 8 -1 User 872524* LOG FULL (246) ! 0 0 27949690(s)

T139 8 -1 User 1081510* LOG FULL (246) ! 0 0 27949690(s)

T140 8 -1 User 782344* LOG FULL (246) ! 0 0 27949690(s)

T141 8 -1 User 765964* LOG FULL (246) ! 0 0 27949690(s)

T142 8 -1 User 622660* LOG FULL (246) ! 0 0 27949690(s)

T143 8 -1 User 979134* LOG FULL (246) ! 0 0 27949690(s)

T144 8 -1 User 938038* LOG FULL (246) ! 0 0 27949690(s)

T145 8 -1 User 446504* LOG FULL (246) ! 0 0 27949690(s)

T146 8 -1 User 1220814* LOG FULL (246) ! 0 0 27949690(s)

T147 8 -1 User 1196114* LOG FULL (246) ! 0 0 27949690(s)

T148 8 -1 User 1216664* LOG FULL (246) ! 0 0 27949690(s)

T149 8 -1 User 462920* LOG FULL (246) ! 0 0 27949690(s)

T150 8 -1 User 839890* LOG FULL (246) ! 0 0 27949690(s)

T151 8 -1 User 856288* LOG FULL (246) ! 0 0 27949690(s)

T152 8 -1 User 1093794* LOG FULL (246) ! 0 0 27949690(s)

T153 8 -1 User 634942* LOG FULL (246) ! 0 0 27949690(s)

T154 8 -1 User 1134844* LOG FULL (246) ! 0 0 27949690(s)

T155 8 -1 User 1044668* LOG FULL (246) ! 0 0 27949690(s)

T156 8 -1 User 688214* LOG FULL (246) ! 0 0 27949690(s)

T157 8 -1 User 1179696* LOG FULL (246) ! 0 0 27949690(s)

T158 8 -1 User 1175766* LOG FULL (246) ! 0 0 27949690(s)

T159 8 -1 User 495680* LOG FULL (246) ! 0 0 27949690(s)

T160 8 -1 User 1142966* LOG FULL (246) ! 0 0 27949690(s)

T161 8 -1 User 901190* LOG FULL (246) ! 0 0 27949690(s)

T162 8 -1 User 389286* LOG FULL (246) ! 0 0 27949690(s)

dwic694:sqdqn2 3>

Thanks for any help you can provide.

lbreddemann
Active Contributor
0 Kudos

Hi Clint,

I don't think that this problem is as bad as you think.

I believe you think it's bad, because all session had been flagged to get killed - sorry, but as MaxDB does not have a preemtive internal task scheduling this will only work as soon as the LOGFULL situation is gone. Ironically, by killing the sessions it now does not make any difference anymore, if you just stop the database or perform a data backup first.

The sessions will be gone afterwards in any case.

Anyhow - here's what would usually be the 'solution' to the 'problem':

You see the LOGFULL situation and login to the database via dbmcli.

Now, when you type in "backup_start ..." you get an error like "-24991,ERR_NODBSESSION: no database session available".

Anyhow, this message does not mean that you cannot take a backup anymore.

All you've to do is to get a 'db session' or a 'utility session' (but not a 'sql session').

Therefore, go ahead and type in (when logged on in dbmcli as 'control'):

> util_connect

> backup_start <medium_name> DATA

Now the backup will be taken and afterwards you can perform a manual log backup or switch on the automatic log backup.

A completely different thing you should think about is, that you still use the sqd<user> although it has been given up from MaxDB 7.5 onwards.

Nowadays the MaxDB software is owned by 'sdb:sdba' and the <sid>adm user is member of the 'sdba' group.

And, yes - please install the current patch of MaxDB. 7.6 Build 18 is so outdated!

best regards,

Lars

former_member229109
Active Contributor
0 Kudos

Hello Clint,

-> You have quit old database version

-> What is the value of the database parameter _MAXEVENTTASKS

"Maximum number of event tasks

The database system uses event tasks both for Event Dispatcher and for the DBM

commands event_wait, event_receive, event_available, auto_extend and auto_update_statistics.

These commands can be active simultaneously for this database instance.

The default value of the Database Manager for this parameter is 2."

-> Please update the ticket with output of the commands:

dbmcli -s db_enum

sdbregview -l

xinstinfo QN2

dbmcli -d <database_name> -u <dbm_user>,<password> dbm_configget RunEventDispatcher

dbmcli -d <database_name> -u <dbm_user>,<password> param_directget _MAXEVENTTASKS

dbmcli -d <database_name> -u <dbm_user>,<password> auto_extend show

dbmcli -d <database_name> -u <dbm_user>,<password> auto_update_statistics show

dbmcli -d <database_name> -u <dbm_user>,<password> event_list

dbmevtdisp version

dbmevtdisp state -d <database_name> -u <dbm_user>,<password>

< You can stop the Event Dispatcher using dbmevtdisp stop >

-> If auto_update_statistics & auto_extend are on, could you please deactivate them & see if you could create the complete databackup & log backup after that.

Thank you & best regards, Natalia Khlopina