5

I am facing a sudden bug that arose from a custom scheduled batch class in the system. This scheduled job has been running smoothly for over a year now, but it seems as though a recent influx of records (~2,000) for which the batch class iterates over has thrown the process over the CPU time limit.

Every time the job runs, the ending "Status" (Setup > Monitor > Jobs > Apex Jobs) is the same as it always has been - "Completed". However, in the "Status Detail" section, there is now a "First error: Apex CPU time limit" message that appears after every run.

After researching potential solutions, I have tried modifying the batch class to avoid this timeout by:

  1. reducing the batch size (down to as low as 10)
  2. modifying the initial query (Start method) - removed sub-fields, etc.
  3. breaking apart/moving around the data processing operations within the Execute method

Nothing has seemed to help at all. When looking at similar posts around the topic, it seems as though the problem lies in the initial query. I've used the Query Plan tool (developer console) to try and optimize/debug the initial query, and it tells me that the two filtering fields are "Not considering filter for optimization because unindexed". Both fields are custom picklist fields.

Am I going down the right path with this? If Salesforce support could make both fields "custom index" fields, would this likely solve the problem / speed things up (i.e. not timeout)? Are there other things/places I should be looking at?

Any help would be greatly appreciated...

Here is the query from the Start method:

return Database.getQueryLocator(
        'SELECT Id, Lookup_1__c, Lookup_2__c, Lookup_1__r.Name, Lookup_2__r.Name, External_Id_Field__c, '+
        'Field_A__c, Name, Lookup_3__c, CreatedDate '+
        'FROM Custom_Object__c '+
        'WHERE Custom_Picklist_Field_1__c = \'Value A\' '+
        'AND Custom_Picklist_Field_2__c = \'Value B\'');
5
  • You should be able to confirm that the problem is in the start not the execute by turning on the debug logs: separate logs generate for start and (multiple) executes (and finish). You can also use the Query Plan Tool to better understand the query and so check that indexing will help.
    – Keith C
    Aug 25, 2016 at 23:36
  • Try and capture the Debug log when the batch is running. This can then be used to find where the CPU time is going. What is the batch job doing in each execution? Aug 25, 2016 at 23:37
  • do you have the option of adding a date filter (like records in the current year) to the query?
    – cropredy
    Aug 26, 2016 at 1:04
  • Unfortunately, I do not have the option of adding a date filter... I've captured time stats around each area of the batch class, and specifically, the "execute" pieces are running anywhere between 50-85 ms. My understanding is that the time limits are at 10,000 ms? The batch class seems to get through all of its chunks, as there are logs for each, and the total time after the last chunk was only at 2,100 ms.
    – zwizard
    Aug 26, 2016 at 15:09
  • Also, I have just noticed that the "timeout" error is shown in the "Finish" batch log every time (i.e. the 2nd batch log with an "Operation" label of "Batch Apex"). When I comment out the processing in that section, the job completes just fine. I think this is obviously the source of the timeout issue. Thanks for the input!
    – zwizard
    Aug 26, 2016 at 15:12

1 Answer 1

1

I think there a few post about this in SFSE, but in summary try to checkout the following :

Make sure that your query is selective. I recommend you to have a read at Salesforce article here. But mainly you need to use index fields, a field will be index if :

If the filter is on a standard field, it'll have an index if it is a primary key (Id, Name, OwnerId), a foreign key (CreatedById, LastModifiedById, lookup, master-detail relationship), and an audit field (CreatedDate, SystemModstamp).

Custom fields will have an index if they have been marked as Unique or External Id

Purge your recycle bin, if you regularly delete records for that particular object, Salesforce take that in consideration, you can do that in apex if required using Database.emptyRecycleBin()

If your query is indexed, make sure it doesn't exceed the threshold. For standard object, it's 30% of the first million and for custom object it's 10% of the first million.

Make sure you don't schedule too many complex batch on the same object at the same time.

If your batch is failing randomly it might also be because Salesforce is having a downtime on their server just at the time your batch is running. Also if their server reset for any reason at that time, your apex job will stop.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .