Question

CPU is randomly spiking to 100% in Task Manager for all-in-one server (LFS, WF & SQL)

Laserfiche Workflow

Updated June 27, 2016

asked on February 2, 2015 • Show version history

I see sporatic CPU spikes, after talking to PreSales, they gave me the counters that I would run in Perfmon, I ran Perfmon, i see ave disk sec/Read at average 0.008, i see avg disk sec/write at average 0.005, Log Flush Waits/sec is averaging 2.014, when i was talking with PreSales, it should be close to 0-1 as possible, Page life expectancy average is 1,688,594. Looking at the MDF vs LDF, LDF is very small in comparison with MDF. The SQL DB location & tempdb files are on a location with enough free space.

Below are the counter results of Perfmon:

Log Flush Waits/sec --> Average: 2.014

Log Flush Wait Time --> Average: 73.027

Lazy writes/sec --> Average: 0

Page life expectancy --> Average: 1,688,594

Avg. Disk sec/Read (D: drive where all mdf & ldf are stored) --> Average: 0.008

Avg. Disk sec/Write (D: drive where all mdf & ldf are stored) --> Average: 0.005

% Processor Time (WF instance) --> Average: 0.694

% Processor Time (LFS instance) --> Average: 0.838

% Processor Time (sqlsrvr instance) --> Average: 1.866

Processor Queue Length --> Average: 0.712

% Processor Time (Total) --> Average: 10.694

% Disk Time (Total) --> Average: 16.493

Page Reads/sec (Memory) --> Average: 12.667

Pages/sec (Memory) --> Average: 158.917

Page Writes/sec (Memory) --> Average: 0.008

So in conclusion, based on these results from Perfmon, 2 things needs to be done:

1. move the sql db to a faster HDD

2. assign more memory to LFS

Is there anything i missed here? Any other recommendations?

0 0

Replies

replied on February 2, 2015

When your CPU spikes what is the memory usage at that time. Is it close to or at 100% as well?

How many processors do you have on the server?

0 0

replied on February 2, 2015

12.83GB out of 15GB

4 processors

0 0

replied on February 2, 2015

In my experience, since the memory is not maxed adding more memory is not going to help you at all. I have a similar situation in the past where the Laserfiche server had one processor and 16 Gigs of memory. It was still maxing the CPU and the memory usage was at about 6 Gigs. All this memory had been poured into the machine but was sitting there unused while the processor was maxed. I would suggest upping the processor cores if possible. I increased my server's processors recently to 8 and since then the complaints about Laserfiche being slow have completely disappeared.

0 0

replied on February 4, 2015

The only counters that jump out at me are the ones that pertain to log flushes and paging--both seem high. That being said, for a lot of those counters, the average isn't going to tell you a ton; if the CPU spends half of its time pegged out at 100% and half of its time at 10%, the average is going to say 55%. Are there any trends in when the CPU is under great strain? Does CPU strain and memory strain correlate with system usage? Does it show up when the Laserfiche system is being used heavily? Are there any events in the event log that match up with the system load?

0 0

replied on February 5, 2015

its a random CPU spikes, so it is hard to really catch the time and see the event logs at that specific time. I do not see any trends when the CPU is spiking, or maybe im not looking at the right place?

0 0

replied on April 10, 2015

Levi,

I just recently had an issue similar to yours. Are you still having the issue?

0 0

View 1 previous reply

replied on June 21, 2016

How many processors did you end up giving to the SQL box?

0 0

replied on June 21, 2016 • Show version history

If you have not specifically told SQL what resources it has, it will use as much as it can get. If that is the case, log into Management Studio, right click the server node and select properties.

First, select Memory on the left side and review the settings there. You can set a minimum and maximum value.

Second, select Processors on the left side and review the settings there.

It is usually preferred to put SQL Server on it's own server because when the software needs to communicate with it, you have more than one program trying to execute tasks at the same time

1 0

replied on June 27, 2016

Levi, hope you could let me know how many CPU's you used after? I boosted mine from 8 to 16 last night and still seeing the same spikes just now on all 16 processors.

0 0

You are not allowed to follow up in this post.

Question

Question

CPU is randomly spiking to 100% in Task Manager for all-in-one server (LFS, WF & SQL)

Replies

Sign in to reply to this post.