High CPU Usage for es_webreports_svc.exe

We just recently started using Event Sentry and I noticed the Web Reporting Service process is suddenly using a lot of CPU. I just updated to the latest version of Event Sentry (3.1.1.17) and the problem is still there. I tried stopping and restarting the service to no avail.

When the service first starts it jumps up to about 25% CPU usage and fluctuates +/- 3% or so. After running for a few minutes, it jumps up to 50% and fluctuates +/- 3% or so. If I then visit the Jobs page at http://127.0.0.1:8080/jobs, the process jumps to at least 75% and sometimes goes as high as 90%, effectively killing the machine. Also, the jobs page itself never finishes loading.

I just recently created a report job to email a pdf version of a report on a monthly basis. This was the first job report I created. Could this be the culprit? What is the expected idle CPU usage for es_webreports_svc.exe?

The machine is a quad-core, Windows Server 2008 R2 SP1, 64-bit machine with an Intel Core i3-2100 processor running at 3.1 GHz; it has 4 GB of RAM. If I can provide further information (debug logs, configuration files, etc.), please let me know.

Thanks,
Mike

Comments

  • Hi Mike,

    Have you had this problem in the past? Unfortunately, there were some known issues with the jobs in the past causing an endless loop when trying to determine the next run time. Could you please rename "JOBS.XML" file and then restart the web reports. Does this correct the issue with the CPU?

    If you then create a new job and schedule it, does the CPU remain at a normal level or does it suddenly use a lot of CPU again?

    If you had a lot of reports that you need help re-creating, we'd be happy to help you with that since it was a problem with the web reports. If you would like help with this, please call our support number and reference this forum post.

    Steven
  • Renaming Jobs.XML and restarting the web reports service fixed the issue with the CPU.

    I then tried recreating the report job (luckily I only have one so far). I did this through the "Edit Report" dialog. I checked the box for "Create Job" then filled in the information to create the job.

    When I click the "Save" button on this dialog, the web browser loading icon begins running continuously (I'm using Google Chrome). I checked the CPU usage on the server and it jumped to 25% for a few minutes, then jumped again to 50%. It seemed to settle in at 50%.

    The web page never did respond, so I clicked on the "Cancel" button. This got rid of the dialog, but the browser loading icon continued to run. The CPU usage stayed at 50%.

    I checked the conf folder, and there was a new Jobs.xml file.

    I then clicked on the Reports --> Jobs link from the web interface and the CPU usage jumped up to 75% when it tried loading the /jobs/ page.
  • edited February 2015
    I shut down the service and restarted it, leaving the Jobs.xml file in place. CPU usage immediately spiked to 25%, leveled off for a few minutes, then jumped up to 50%. As I said earlier, this is a quad core processor. My best guess is that one of the threads is using all of the available CPU, then a new thread is created (or the processor moves an existing one to a different core) which also uses all of the available CPU time.

    I'll post the full contents (with obfuscated email addresses) of the Jobs.xml file here. I don't see anything in the file that looks bizarre, but perhaps it will help you in your troubleshooting:

    <?xml version="1.0" encoding="UTF-8" standalone="yes"?>
    <jobs version="1" autoIncrement="1">
    <jobs>
    <entry>
    <key>1</key>
    <value>
    <bcc></bcc>
    <cc>xxx@gmail.com</cc>
    <created>2015-02-27T15:35:26.628Z</created>
    <enabled>true</enabled>
    <frequency>monthly</frequency>
    <message>The attached report lists the changes to the Active Directory security groups over the past month.</message>
    <monthly>
    <days>Last</days>
    <months>all</months>
    <option>days</option>
    </monthly>
    <name>Active Directory Security Group Change Recap</name>
    <priority>3</priority>
    <recipients>xxx@ourdomain.com</recipients>
    <report>2</report>
    <reportFormat>pdf</reportFormat>
    <sendEmpty>true</sendEmpty>
    <senderEmail>xxx@ourdomain.com</senderEmail>
    <start>2015-02-27T23:30:00Z</start>
    <subject></subject>
    <type>report</type>
    </value>
    </entry>
    </jobs>
    </jobs>
  • I found the problem!

    It's with the "Last" option in days when we set up a monthly schedule:
    <monthly>
    <days>Last</days>
    <months>all</months>
    <option>days</option>
    </monthly>
    If I change the option from "Last" to a number (e.g., "1") then the CPU problems go away.

    I'm guessing there is a problem parsing the file where the "Days" value is assumed to be a number and it breaks when it finds text. "Last" is one of the options in the web interface drop-down when setting up a monthly report.

    For now, I will just switching to using a number date, but being able to set up a report to run on the last day of each month would be handy.
  • Hi Mike,

    Thanks for finding the issue, we'll pass this on to the web development team and see that it will get resolved asap.

    I expect that we'll have a fix available mid next week; if you send an email to support then we can let you know when it's available.

    Thanks again!
Sign In or Register to comment.