Picture Perfect 4.6 Enterprise Edition User Manual - UTCFS Global ...

Picture Perfect 4.6 Enterprise Edition User Manual - UTCFS Global ... Picture Perfect 4.6 Enterprise Edition User Manual - UTCFS Global ...

utcfssecurityproducts.com
from utcfssecurityproducts.com More from this publisher
13.07.2015 Views

90Picture Perfect 4.6 Enterprise EditionUser ManualNetwork outage on one hostIf a host is down for any reason, all changes destined for that host remain in queue. This means thatall changes are automatically applied when the host is restarted or brought back into the network.There is a limit to how long the host can be disconnected and yet have all its changes queued.The only limitations are the size of the queue (32 MB) and the size of the ER database spaces (whichare set to 40% of the base database sizes). Once the database spaces get filled to about 90%, analarm is raised and the host needs to be removed from the ER configuration using ercmd.sh.Based on our testing, the time a host can be disconnected and yet not have to be disconnected is inthe order of weeks.Note:This is a very rough estimation and depends a lot on the sizing of the database spaces and the number ofchanges being made.With an ER database space size of 200 MB, about 500K transactions can be made before 90%space is filled up. With 100 GB of ER database space, about 250M transactions can be made (giventhese transactions would have to be non-history related tables). It is very unlikely a host would haveto be removed from the ER, possibly for months.A customer can extend this time, by simply extending the size of the database space usingercmd.sh.Details of differences reported by ercmd.sh --statusUse cdr check with the -verbose option. See Informix Help for more information on cdr check.ATS/RIS alarmsIf ATS/RIS alarms are reported, check the /ppbackup/ris and /ppbackup/rts directories forthe files reporting the error(s). Depending on the error reported, a 'repair' command can be run fromthe nethost OR the system administrator may be able to decide which data should be 'winning' andtake action accordingly.Example of an ATS error that can be repaired by a cdr command:ATS output:root@bcteaglelake# cat ats.subhost_6.subhost_3.D_50.110531_14:40:28.1TXH Source ID:3 / Name:subhost_3 / CommitTime:11-05-31 14:40:27TXH Target ID:6 / Name:subhost_6 / ReceiveTime:11-05-31 14:40:28TXH Number of rows processed when transaction was aborted:1TXH Error: Dead lock encounteredTXH CDR:0 / SQL:-244 / ISAM:-143----------RRH Row:1 / Replicate Id: 65538 / Table: proteus@informix.badge / DbOp:DeleteRRS 3 (subhost_3)|1306867227 (11/05/31 14:40:27)RRD 268435458|Badge in bcthouston|0268435458||268435458||||||||||1|||||-1|20110531|144027In this case the deletion failed on subhost_3 because the badge table was in use (locked). To repairthis, run the following command on the nethost.root@bctcali# su - informix -c "cdr view -c subhost_6 atsdir -R -v"

Chapter 6Troubleshooting and support91ATSDIRServer File Size CreateNameTime---------------------------------------------------------------------------subhost_6 ats.subhost_6.subhost_3.D_50.110531_14:40:28.1 483 2011-05-3114:40:28Attempting connection to syscdr@bctcali...Using syscdr@bctcali.Source ID:3 / Name:subhost_3Target ID:6 / Name:subhost_6(1) [bid = "0268435458"]: Row not found on the source for replicate (1) [bid = "0268435458"]: Row will be deleted on the target for replicateSample output of ATS/RIS files for errors that require the system administrator to take action:root@bctlesabre3# cat /ppbackup/ats/ats.nethost_1.subhost_3.D_104.110510_01:3TXH RIS file:/ppbackup/ris/ris.nethost_1.subhost_3.D_104.110510_01:39:23.1 hasalso been created for this transaction==========TXH Source ID:3 / Name:subhost_3 / CommitTime:11-05-10 01:39:23TXH Target ID:1 / Name:nethost_1 / ReceiveTime:11-05-10 01:39:23TXH Number of rows processed when transaction was aborted:1TXH All rows in a transaction defined with row scope were rejected----------RRH Row:1 / Replicate Id: 65547 / Table: proteus@informix.host_bid_format /DbOp:UpdateRRS 3 (subhost_3)|1305005963 (11/05/10 01:39:23)RRD 3|Facility2 String10|00%10s|-1|20040202|161751root@bctlesabre3# cat ris.nethost_1.subhost_3.D_104.110510_01:39:23.1TXH Source ID:3 / Name:subhost_3 / CommitTime:11-05-10 01:39:23TXH Target ID:1 / Name:nethost_1 / ReceiveTime:11-05-10 01:39:23----------RRH Row:1 / Replicate Id: 65547 / Table: proteus@informix.host_bid_format /DbOp:UpdateRRH CDR:0 / SQL:-239 / ISAM:-100RRS 3 (subhost_3)|1305005963 (11/05/10 01:39:23)RRD 3|Facility2 String10|00%10s|-1|20040202|161751==========TXH Transaction abortedTXH ATS file:/ppbackup/ats/ats.nethost_1.subhost_3.D_104.110510_01:39:23.2 hasalso been created for this transactionIn the above example, the reason for the failure to insert a record from subhost to nethost is thatthere was a unique index violation. Depending on whether this is the primary key or not, the actionrequired would be different.• If this is the primary key, then delete the records from both hosts and recreate the record fromone of the hosts.• If the unique index violation is not on the primary key, then delete the record from one of thehosts (this decision is based on data in other fields of the record) and let the other recordreplicate.If the error cannot be handled, please contact Technical Support.

90<strong>Picture</strong> <strong>Perfect</strong> <strong>4.6</strong> <strong>Enterprise</strong> <strong>Edition</strong><strong>User</strong> <strong>Manual</strong>Network outage on one hostIf a host is down for any reason, all changes destined for that host remain in queue. This means thatall changes are automatically applied when the host is restarted or brought back into the network.There is a limit to how long the host can be disconnected and yet have all its changes queued.The only limitations are the size of the queue (32 MB) and the size of the ER database spaces (whichare set to 40% of the base database sizes). Once the database spaces get filled to about 90%, analarm is raised and the host needs to be removed from the ER configuration using ercmd.sh.Based on our testing, the time a host can be disconnected and yet not have to be disconnected is inthe order of weeks.Note:This is a very rough estimation and depends a lot on the sizing of the database spaces and the number ofchanges being made.With an ER database space size of 200 MB, about 500K transactions can be made before 90%space is filled up. With 100 GB of ER database space, about 250M transactions can be made (giventhese transactions would have to be non-history related tables). It is very unlikely a host would haveto be removed from the ER, possibly for months.A customer can extend this time, by simply extending the size of the database space usingercmd.sh.Details of differences reported by ercmd.sh --statusUse cdr check with the -verbose option. See Informix Help for more information on cdr check.ATS/RIS alarmsIf ATS/RIS alarms are reported, check the /ppbackup/ris and /ppbackup/rts directories forthe files reporting the error(s). Depending on the error reported, a 'repair' command can be run fromthe nethost OR the system administrator may be able to decide which data should be 'winning' andtake action accordingly.Example of an ATS error that can be repaired by a cdr command:ATS output:root@bcteaglelake# cat ats.subhost_6.subhost_3.D_50.110531_14:40:28.1TXH Source ID:3 / Name:subhost_3 / CommitTime:11-05-31 14:40:27TXH Target ID:6 / Name:subhost_6 / ReceiveTime:11-05-31 14:40:28TXH Number of rows processed when transaction was aborted:1TXH Error: Dead lock encounteredTXH CDR:0 / SQL:-244 / ISAM:-143----------RRH Row:1 / Replicate Id: 65538 / Table: proteus@informix.badge / DbOp:DeleteRRS 3 (subhost_3)|1306867227 (11/05/31 14:40:27)RRD 268435458|Badge in bcthouston|0268435458||268435458||||||||||1|||||-1|20110531|144027In this case the deletion failed on subhost_3 because the badge table was in use (locked). To repairthis, run the following command on the nethost.root@bctcali# su - informix -c "cdr view -c subhost_6 atsdir -R -v"

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!