Quantcast
Channel: SCN: Message List
Viewing all articles
Browse latest Browse all 2695

Re: waiting for regular buffer read to complete during MOUNT DATABASE

$
0
0

Assuming a largish database and slow NFS access ... 100 seconds (100,000 ms) doesn't sound like a lot of unexpected wait time.  Would really need a good bit more info ...

 

- Has this same exact mount (same NFS files, same ASE and config) run faster in the past? does it run fast sometimes but slow at other times? [assuming *NO* changes by you ... may want to speak to your network folks to see if you're competing against processing that sporadically floods your network or NFS host with activity]

 

- How big is the database, are all devices on NFS, and what was the total time to get the database mounted and online for use?

 

- Have you tried wrapping your 'mount database' in some sampling of metrics to get a better picture of what's going on? For example:

 

================= test harness

select 'TimeStamp' = convert(varchar,getdate(),109)

select * from master..monDeviceIO

select * from master..monNetworkIO

select * from master..monProcessActivity where SPID = @@spid

select * from master..monProcessObject where SPID = @@spid

select * from master..monProcessStatement where SPID = @@spid

 

select * from master..monProcessWaits where SPID = @@spid

go

exec sp_sysmon "begin_sample"

go

 

mount database ...

go

 

select 'TimeStamp' = convert(varchar,getdate(),109)

select * from master..monDeviceIO

select * from master..monNetworkIO

select * from master..monProcessActivity where SPID = @@spid

select * from master..monProcessObject where SPID = @@spid

select * from master..monProcessStatement where SPID = @@spid

select * from master..monProcessWaits where SPID = @@spid

go

exec sp_sysmon "end_sample",noclear

go

=================

 

... yeah, there's going to be some redundant info in all of that output, but I'd rather have too much than not enough.  And (obviously) it's probably best to run your test with little/no other activity so that you're not skewing your samples.


Viewing all articles
Browse latest Browse all 2695

Trending Articles