SDS_ENABLE is not set error after server restart

Hi all, Informix 11.10 We are moving to new servers and while I have time I wanted to set up an HDR pair to test failover back and forth before I put these into production. From our existing primary, I did a level 0 dump over the wire to the new server. ontape -s -L 0 -F -t STDIO | ssh 192.168.100.229 ". /etc/profile.d/cargotel.sh; ontape -r -t STDIO" Then I brought the server up as a standalone with onmode -m That succeeded. I then shut the server down with onmode -ky Tried to bring it up with oninit which fails with this error: [informix@746286-ORDDBI1 ~]$ Can not initialize SDS - SDS_ENABLE is not set, but the rootchunk indicates a shared disk environment The online.log has the following: 10:59:35 IBM Informix Dynamic Server Stopped. 10:59:53 IBM Informix Dynamic Server Started. 10:59:53 Segment locked: addr=0x44000000, size=114050392064 10:59:53 Segment locked: addr=0x1ad1eee000, size=774144000 Mon Dec 21 10:59:55 2015 10:59:55 pid 13445: can't get real path of 10:59:55 pid 13445: scan_eh_frame(): bad CIE reference 11:00:05 Event alarms enabled. ALARMPROG = '/opt/IBM/informix/etc/alarmprogram.sh' 11:00:05 Booting Language <c> from module <> 11:00:05 Loading Module <CNULL> 11:00:05 Booting Language <builtin> from module <> 11:00:05 Loading Module <BUILTINNULL> 11:00:13 DR: DRAUTO is 0 (Off) 11:00:13 DR: ENCRYPT_HDR is 0 (HDR encryption Disabled) 11:00:13 Event notification facility epoll enabled. 11:00:14 IBM Informix Dynamic Server Version 11.10.FC2 Software Serial Number AAA#B000000 11:00:18 Can not initialize SDS - SDS_ENABLE is not set, but the rootchunk indicates a shared disk environment 11:00:18 IBM Informix Dynamic Server Stopped. 11:00:18 mt_shm_remove: WARNING: may not have removed all/correct segments Our production server does not have SDS_ENABLE set in it's onconfig. I'm told we have had this problem in the past and that it is because we are trying to bring this up as an HDR primary and the workaround we used before was to bring it up as an HDR secondary without connecting it to the primary. That sort of makes sense but I don't know how to do that, or indeed if that's the proper solution. Thanks you, Jeff Ross

Original post: Hi all, Informix 11.10 We are moving to new servers and while I have time I wanted to set up an HDR pair to test failover back and forth before I put these into production. From our existing primary, I did a level 0 dump over the wire to the new server. ontape -s -L 0 -F -t STDIO | ssh 192.168.100.229 ". /etc/profile.d/cargotel.sh; ontape -r -t STDIO" Then I brought the server up as a standalone with onmode -m That succeeded. I then shut the server down with onmode -ky Tried to bring it up with oninit which fails with this error: [informix@746286-ORDDBI1 ~]$ Can not initialize SDS - SDS_ENABLE is not set, but the rootchunk indicates a shared disk environment < ... stuff removed ...> 11:00:18 Can not initialize SDS - SDS_ENABLE is not set, but the rootchunk indicates a shared disk environment 11:00:18 IBM Informix Dynamic Server Stopped. 11:00:18 mt_shm_remove: WARNING: may not have removed all/correct segments Our production server does not have SDS_ENABLE set in it's onconfig. I'm told we have had this problem in the past and that it is because we are trying to bring this up as an HDR primary and the workaround we used before was to bring it up as an HDR secondary without connecting it to the primary. That sort of makes sense but I don't know how to do that, or indeed if that's the proper solution. Thanks you, Jeff Ross Response: I'm curious what output (if any) do you get from onstat -g sds verbose on the production machine you took the archive from? Also, what output do you get from oncheck -pr output from that instance, only interested in the output for "Validating PAGE_PZERO"...so that output would look something like this: Validating IBM Informix Dynamic Server reserved pages Validating PAGE_PZERO... Identity IBM Informix Dynamic Ser ver Copyright 2001, 2015 IBM Corporation Database system state 0 Database system flags 0x9 64-bit server BigChunk page flags are not in use Page Size 2048 (b) Date/Time created 12/04/2015 10:45:27 Version number of creator 28 Last modified time stamp 0 UID of rootdbs creator 200 Index Page Logging ON since 12/04/2015 10:45:27 HA Disk Owner <null> Validating PAGE_CONFIG... The output would continue on, but I'm only interested in the stuff between the PAGE_PZERO and the PAGE_CONFIG. Particularly the "HA Disk Owner" field. Jacques Renaut IBM Informix Advanced Support APD Team

Hi and thanks for the follow-up! Here are the results you asked for: onstat -g sds verbose IBM Informix Dynamic Server Version 11.10.FC2 -- On-Line (Prim) -- Up 151 days 19:08:06 -- 116497336 Kbytes Number of SDS servers:0 Updater node alias name :chisec oncheck -pr Validating IBM Informix Dynamic Server reserved pages Validating PAGE_PZERO... Identity IBM Informix Dynamic Ser ver Copyright(C) 1986-20 07 IBM Informix Softwar Database system state 0 Database system flags 0x3 Page Size 2048 (b) Date/Time created 01/31/2008 05:29:16 Version number of creator 15 Last modified time stamp 0 Index Page Logging ON since 05/11/2010 14:33:06 HA Disk Owner chisec

Still on Informix 11.10.FC2 I've just again encountered the same error while attempting to restore our database from a level 0 backup to a new development server: I scp'ed the backup file from our normal daily backups to the dev server and restored it with cat CGTInformixBackup.20160212.Thu.L0 | ontape -r -t STDIO This succeeded. I then brought the server on-line with onmode -m and that succeeded. Brought the server down with onmode -ky Tried to bring it back on-line with onint and got the same error message: 10:12:35 IBM Informix Dynamic Server Stopped. 10:12:48 IBM Informix Dynamic Server Started. 10:12:48 Segment locked: addr=0x44000000, size=114050392064 10:12:48 Segment locked: addr=0x1ad1eee000, size=2871296000 Tue Feb 16 10:12:50 2016 10:12:50 pid 13978: can't get real path of 10:12:50 pid 13978: scan_eh_frame(): bad CIE reference 10:13:03 Event alarms enabled. ALARMPROG = '/opt/IBM/informix/etc/alarmprogram.sh' 10:13:03 Booting Language <c> from module <> 10:13:03 Loading Module <CNULL> 10:13:03 Booting Language <builtin> from module <> 10:13:03 Loading Module <BUILTINNULL> 10:13:11 DR: DRAUTO is 0 (Off) 10:13:11 DR: ENCRYPT_HDR is 0 (HDR encryption Disabled) 10:13:11 Event notification facility epoll enabled. 10:13:12 IBM Informix Dynamic Server Version 11.10.FC2 Software Serial Number AAA#B000000 10:13:12 Can not initialize SDS - SDS_ENABLE is not set, but the rootchunk indicates a shared disk environment 10:13:12 IBM Informix Dynamic Server Stopped. 10:13:12 mt_shm_remove: WARNING: may not have removed all/correct segments Last time I ended up adding the new servers in as RSS, then switching them to HDR secondary. I want a standalone database in the case so I'll try to do the RSS and then switch to standalone. The bigger issue at hand, though, is I have not been able to successfully restore an informix database from a level 0 backup and that scares me. I'm taking these backups on the production server with: ontape -v -s -L 0 -t STDIO | ssh $REMOTE_SERVER " gzip > $REMOTE_STORE/$BACKUP_FILENAME.gz" I'd sure appreciate a sanity check on this--if we had a true catastrophic failure I'm not convinced I could successfully restore the database and have it remain accessible past the first stop/restart cycle. Thanks! Jeff Ross