Welcome to the SRP Forum! Please refer to the SRP Forum FAQ post if you have any questions regarding how the forum works.

Request limits

Are there any limitations on the number of requests that the engine server can queue?
I'm finding not all the expected requests are completing. Some are. Some aren't. That's for requests of the same type. I'll build in extensive logging if necessary but I'm just looking to eliminate/identify some possibilities first.

Comments

  • There are no imposed limits, but there could be resource limits. We have some sites that hold up rather well under constant extreme loads. Can you give me an estimate of how many requests you are pushing through when you see these problems? How many engines are you running?
  • All I can answer at this stage is two engines.
    I couldn't even provide a reliable estimate as it's too new and not yet logged.
    It's also being used for two separate functions.
    1. to keep an ecommerce site in sync whenever stock/prices change
    2. to keep cloud reporting data current. This is the newest addition and the one I've noticed something amiss. Nothing in July had yet been updated. I restarted the engineserver and kicked July updates off manually and it was successful for 210 records after three attempts. By three attempts I mean it incrementally updated. Some new logging on the receiving end showed there were some rate limits hit so that would make sense why some records made it and some didn't.
    I don't think it explains why none of the July records had made it previously.

    I've modified to use a batch approach rather than an individual record approach to alleviate rate limit concerns.
    That might be sufficient or it might be that the ecommerce transactions are queueing up and I'm just seeing the symptoms on the reporting side because that's more obvious when data is missing.

    I'll also add a third engine to the mix and see how things progress.
  • As far as you can tell, does the SRP Engine Server show new activity and do the engines seems like they are processing requests? Are the engines themselves being responsive (i.e., don't show evidence of being hung)? Are you submitting requests from elsewhere via SRP_TcpClient or are you running an autocommand with an interval?
  • It's set up as a windows task so it's difficult to tell if anything is happening.
    There is an auto task running every ten minutes or so, but it should be relatively quick. A simple GET request that likely comes back with no results to process. Everything else is via SRP_TcpClient().

    My suspicion is that the updates to the web might be causing a backlog so I'm just feeling out whether it could be that things just fall exponentially behind or that they reach a point where the engineserver just has too many requests to manage.

    I might just run it from a user profile instead of as a task for a couple of days and have the engines visible so I can eyeball what's happening.
Sign In or Register to comment.