Search the Community

Showing results for tags 'troubleshooting'.

  • Search By Tags

    Type tags separated by commas.
  • Search By Author

Content Type


  • From LogicMonitor
    • General Announcements
    • LM Staff Contributions
    • Community Events
  • LogicMonitor Product Discussion
    • Feature Requests
    • LM Exchange
    • Ask the Community

Find results in...

Find results that contain...

Date Created

  • Start


Last Updated

  • Start


Filter by number of...


  • Start



About Me

Found 4 results

  1. Hi, Per discussion with Russ G. & Kenyon W. & Jake C. yesterday, I would like to submit this as a feature request to the DEV team and see whether there is any way to add this feature into future roadmap. In short, it'll be great if end user can configure multiple incident/alerts into 1 group and generate only 1 alert (with highest severity). Here is an example of Tomcat being shutdown which shows a number of alerts generated: 1. Tomcat shutdown ‘critical’ alert is generated (1 alert) 2. ActiveMQ consumer count of specific queue alert has reached zero ‘Error’ alert (about 10-12 alerts for our case) In this case end user would like to be able to configure such that LM will consolidate all alerts into one critical alert (i.e. all AMQ 'Error' alerts are cleared)? I saw something like this in PagerDuty and must say it’s a great feature to have in LogicMonitor to reduce # of alerts being processed by the TechOps team: Thanks & Best Regards, Horace
  2. This article provides information on High CPU usage on the Collector . (1) General Best Practices (a) First and foremost we advise our customers to be on latest General Release Collectors (unless advised not to) . Further information all the Collector information could be retrieved on the link below : Also on the release notes of each newer Collector version we will indicate if we have fixed any known issues : (b) Please also view our Collector Capacity guide to get a full overview on how to optimise the Collector Performances : (c) When providing information on High CPU usage it would be useful if you can advise if the High CPU usage is all the time or a certain timeframe only (also if any environmental changes were done on physical machine that may have triggered this issue). Please do advise also if this occurred after adding newer devices on the collector or if this issue occurs after applying a certain version of the Collector. (2) Common Issues On this topic i will go through some of the common issues which have been fixed or worked upon by our Development Teams : (A) Check if the CPU is used by the Collector (Java Process) or SBproxy or other processes. (i) To monitor Collector Java Process : Use the datasource Collector JVM status to check the Collector (Java process) CPU usage (as shown below). (ii) To monitor the SBProxy usage : We can use the datasource : WinProcessStats.xml (for Windows collector / For Linux data source (this datasource is still being developed) . (B) If the high CPU usage is caused by the Collector Java processes, below are some of the common causes : (i) Collector java process using high CPU How confirm if this the similar issue : In the Collector Wrapper Logs you are able to view this error message : In our Collector wrapper.log, you can see a lot of logs like the below: DataQueueConsumers$] Un-expected exception - Must be BUG, fix this, CONTEXT=, EXCEPTION=The third long is not valid version - 0 java.lang.IllegalArgumentException: The third long is not valid version - 0 at com.santaba.agent.reporter2.queue.QueueItem$Header.deserialize( at com.santaba.agent.reporter2.queue.impl.QueueItemSerializer.head( This issue has been in Collector version EA 23.200 (ii) CPU load spikes on Linux Collectors As shown in the image below the CPU usage of Collector Java process has a periodic CPU spike (on an hourly basis) . This issue has been fixed on Collector version EA 23.026 (iii) Excessive CPU usage despite not having any devices running on it In the collector wrapper.log, you can see similar logs as below : [04-11 10:32:20.653 EDT] [MSG] [WARN] [pool-20-thread-1::sse.scheduler:sse.scheduler] [SSEChunkConnector.getStreamData:87] Failed to get SSEStreamData, CONTEXT=current=1491921140649(ms), timeout=10000, timeUnit=MILLISECONDS, EXCEPTION=null java.util.concurrent.TimeoutException at java.util.concurrent.FutureTask.get( at com.logicmonitor.common.sse.connector.sseconnector.SSEChunkConnector.getStreamData( at com.logicmonitor.common.sse.processor.ProcessWrapper.doHandshaking( at com.logicmonitor.common.sse.processor.ProcessorDb._addProcessWrapper( at com.logicmonitor.common.sse.processor.ProcessorDb.nextReadyProcessor( at com.logicmonitor.common.sse.scheduler.TaskScheduler$ at java.util.concurrent.Executors$ at at java.util.concurrent.ThreadPoolExecutor.runWorker( at java.util.concurrent.ThreadPoolExecutor$ at This issue has been fixed on EA 24.085 (iv) SSE process stdout and stderr stream not consumed in Windows Please note this issue occurs on only on Windows Collectors and the CPU usage of the Windows operating system has a stair-step shape as shown below. This has been fixed in Collector EA 23.076 (v) Collector goes down intermittently on daily basis In the Collector wrapper.logs, you can see similar log lines : [12-21 13:10:48.661 PST] [MSG] [INFO] [pool-60-thread-1::heartbeat:check:4741] [Heartbeater._printStackTrace:265] Dumping HeartBeatTask stack, CONTEXT=startedAt=1482354646203, stack= Thread-40 BLOCKED ( com.santaba.common.logger.Logger2$1.print ( com.santaba.common.logger.Logger2._log ( com.santaba.common.logger.Logger2._mesg ( ( com.santaba.agent.util.Heartbeater$HeartBeatTask._run ( com.santaba.agent.util.Heartbeater$ ( java.util.concurrent.Executors$ ( ( java.util.concurrent.ThreadPoolExecutor.runWorker ( java.util.concurrent.ThreadPoolExecutor$ ( ( [12-21 13:11:16.597 PST] [MSG] [INFO] [pool-60-thread-1::heartbeat:check:4742] [Heartbeater._printStackTrace:265] Dumping HeartBeatTask stack, CONTEXT=startedAt=1482354647068, stack= Thread-46 RUNNABLE ( com.santaba.common.logger.Logger2$1.print ( com.santaba.common.logger.Logger2._log ( com.santaba.common.logger.Logger2._mesg ( ( com.santaba.agent.util.Heartbeater$HeartBeatTask._run ( com.santaba.agent.util.Heartbeater$ ( java.util.concurrent.Executors$ ( ( gobler terminated ERROR 5296 java.util.concurrent.ThreadPoolExecutor.runWorker ( java.util.concurrent.ThreadPoolExecutor$ ( ( This issue has now been fixed in Collector EA 22.228 (C) High CPU usage caused by SBProxy (i) Collector CPU spikes until 99% The poor performance of WMI or PDH data collection on some cases will cause too many retries will occur and this consumes a lot of CPU. In the collector sbproxy.log, you can search the log string as shown below and you can see the retry times is nearly 100 per request and subsequently this will consume a lot of CPU. ,retry: This is being investigated by our development team at this time and will be fixed in the near future . (3) Steps to take when facing high CPU usage for Collector (i) Ensure the collector has been added as a device and enabled for monitoring : There are set of New Datasources for the Collector (LogicMonitor Collector Monitoring Suite - 24 DataSources) which as shown below and please ensure they have been updated in your portal and applied to your Collectors and also ensure the Linux CPU or Windows CPU datasources have been applied to the Collector : (ii) Record a JFR (java flying record) in debug command window of the Collector : this can done through this method : // unlock commercial feature !jcmd unlockCommercialFeatures // start a jfr , in real troubleshooting case, should increase the duration a reasonable value. !jcmd duration=1m delay=5s filename=test.jfr name=testjfr jfrStart // stop a jfr !jcmd name=testjfr jfrStop // upload the jfr record !uploadlog test.jfr (iii) Upload the Collector Logs : From the Manage dialog you can send your logs to LogicMonitor support. Select the manage gear icon for the desired collector and then select 'Send logs to LogicMonitor': Credits: LogicMonitor Collector development team for providing valuable input in order to publish this article .
  3. One of the most common support cases we face every day is 'why am I receiving this alert', this article would explain to you the steps on how to determine why are you receiving the alerts. 1) Understand the alert received 2)Checking on validity via raw data and threshold 3)Checking on delivery 1) Understanding the alert received The first step when you receive an alert either via email, text or via any ticketing system is to understand the alert. Understand an alert is to look at which device is the alert for, which datapoint and value of the alert. For example in an email alert message, it would appear as per below. LogicMonitor Alert: Host: ##HOST## Host Group: ##GROUP## Datasource: ##DATASOURCE## Datapoint: ##DATAPOINT## Description: ##DSIDESCRIPTION## Value: ##VALUE## Level: ##LEVEL## Start: ##START## Duration: ##DURATION## Reason: ##DATAPOINT## ##THRESHOLD## ##ALERTID## 2) Checking on validity via raw data and threshold Next, once you determined the alert source, you need to understand why this alert is triggered. This can be done by first looking at the threshold that is set for that particular datapoint. After checking the threshold you can go to the raw data tab of the datapoint to check if it meets the threshold being sent. For example In this case, a critical alert was received and a threshold of 80 90 95 and an alert will only be triggered if you have 20 consecutive polls that fall within this range. Now the next step would be to check on the RAW DATA tab to determine if this condition was met. Judging from the raw data above if you look at the values all the 20 polls have met the threshold level of 80 90 95, but to determine the level of the alert it would be the last poll since the last poll was 96.67 will falls to the range of a critical alert thus a critical alert was send. 3) Checking on delivery The last process is to check the alert rule and escalation chain to see if it was applied to the correct rule and escalation chain. To do so you can go the alert tuning tab and check on the alert routing for that particular instance and datapoint. Here you can see that the Alert Rule applied is Critical - Default and the Alert Chain/Escalation Chain is Critical - Default. Under the Alert Chain is the list of email address that will receive a notification, when the threshold is met.
  4. Per discussion with Jeff Woeber, I want to submit this as feature request in LogicMonitor end as each alert threshold within each datasource (e.g. Tomcat ThreadPool- ) can have its own wiki troubleshooting page. It’s be a great feature if LogicMonitor enables user to specify it’s own troubleshooting page as optional field for each datasource. User can customize specific wiki page as recommendation whenever an alert is sent to PagerDuty.