Problems reported during SQL backup windows.

[MarkLFT]

When our MS SQL Server performs backups, both full and transaction log, we get tow different types of problems reported.

Firstly, we get an MSSQL: Too many physical reads occurring. But I may be wrong, but I would expect there to be a lot of physical reads when performing backups. How can we prevent this problem during backups?

Secondly, I get an MSSQL: Service is unavailable. This I would not expect, and from what I can see, the SQL Server is still running normally. This only lasts a short time, but I cannot explain it, and as we perform backups several times per day, this is really quite annoying.

The SQL server is remote, so we are using a proxy. The proxy itself is not reporting any issues.

Is anyone else experiencing these issues, and if so how do you deal with them?

[cyber]

At any case, do not mention your instance version, templates, what you are using etc... we all have crystal balls on the table..
But if you are using for example this templte https://www.zabbix.com/integrations/mssql#mssql_agent2 then theres a macro {$MSSQL.PERCENT_READAHEAD.MAX} which is used for this trigger and you can adjust the value. Just look up, what your data shows for those moments and do not adjust too far away from that..
Or add time restrictions to this trigger, so it will not activate at backup hours (... and time>xxxxxx and time<yyyyyy). Or set a maintenance period for this timeframe, you can do it with or without data collection. But even if collecting data and triggers will fire, escalations will not be done, so emails/tickets what you may have configured, will not be created..

MSSQL: Too many physical reads occurring

If this value makes up even a sizeable minority of the total "Page Reads/sec" (say, greater than 20% of the total page reads), you may have too many physical reads occurring.

last(/MSSQL by Zabbix agent 2/mssql.readahead_pages_sec.rate) > {$MSSQL.PERCENT_READAHEAD.MAX} / 100 * last(/MSSQL by Zabbix agent 2/mssql.page_reads_sec.rate)

Warning

Service is unavailable... That I cannot predict.. trigger is based on connection availability
last(/MSSQL by Zabbix agent 2/net.tcp.service[tcp,{$MSSQL.HOST},{$MSSQL.PORT}])=0

So if it reports this to be missing, then probably something is cut of at some moment.. can be somethig in your environment You can remake that trigger to consider a bit more time to react... like max(/host/item,#3)=0 (3 last values have to be 0 to activate, making one failure OK).