Diagnostics database archiving guide
Irrespective of the repetitive inspection, carried out by archiving job of evweb periodically at every 6 hours, there could be situations where evweb database is stressed due to continuous flow events. To handle such stressed conditions, there is provision to clean the evweb DB by tracking the number of events received since the product startup. Once the number of events inserted in the evweb meets the threshold value of 100, the archiving is triggered automatically for clearing the space. For more information, see Archiving in Evweb under stressed conditions
Archiving in LOGDB/LOGARCHDB database
The archiving is triggered on the initial product
order to trigger archiving. |
|
|
cimprovider | (disabling the SFM) | |
cimprovider | (enabling the SFM) | |
| OR |
|
cimserver | (stopping the cimserver) |
|
cimserver | (starting the cimserver) |
|
As soon as the product restarts/starts, the job responsible for performing archiving in LOGDB registers itself as a repetitive job. Once successfully registered, it gets called after every 72 minutes. Itchecks if archiving is required based on the space and age conditions.
If any of the criteria as defined in the age/space condition are met, the archiving job triggers the archiving and deletes sufficient amount of space from the current and archive DBs.
The formula used for calculating the number of pages:
Pages to freed up = [Actual number of pages currently held] - [configured number of pages] + [15% (configured number of pages)]
To Trigger Archiving in IOProviders
The IO providers do not employ any mechanism for the archiving. The events in the respective database of IO providers use the following to delete the events from the database.
1.On restarting the product, the IO providers checks for the number of events in the DB. If the number of events in DB is more than 4000, then 25% of 4000 i.e., 1000 events are deleted from the database.
2.However, it is worth noting here that, the check for the number of events exceeding more than 4000 events is used only at the product startup.
Archiving in Evweb under stressed conditions
SysFaultMgmt has an internal counter which keeps track of the number of events been inserted in the evweb DB. When this counter exceeds 100, archiving is called to empty the space by either transferring from the current DB to archive DB or by deleting the events from the archive DB. Events from archive DB are deleted if the number of pages currently held by the archive DB exceeds configured limits.
Limitation of Archiving Mechanism
The archiving mechanism in the SysFaultMgmt deletes events in order to free up the space. The Postgres vacuum daemon refreshes the meta tables related to statistics regularly. But it is seen that the heavily accessed (insert, select, update) DBs, often requires VACCUUM FULL operation to be called on them to update the internal
6