That doesn't add a hard limit to the maximum number of instances that GAE can spin up, it's just a hint to the scheduler that it shouldn't spin up a new instance unless the current incoming request will have to wait longer than 10s to process with existing instances. If GAE is playing up, it's very likely that an incoming request *will *have to wait this long and boom, you've got runaway instances being spun up.
At the moment you have no way to limit the instances - if the scheduler went completely awry, they could keep spinning up instances until the cows come home and you have no way to stop it. On Wednesday, 13 June 2012 21:20:16 UTC+1, alex wrote: > > Simon, so you're saying that setting Min Pending Latency slider to e.g. > 10s does not work for you? > > > On Wednesday, June 13, 2012 10:09:20 PM UTC+2, Simon Knott wrote: >> >> I still find it bizarre that we can't cap the maximum number of >> instances. >> >> If we could say "I never want any more than 5 instances", then this >> billing problem goes away - sure, your service will probably be hit with >> performance issues, but at least you are in more control of your outgoing >> costs. >> >> On Wednesday, 13 June 2012 09:20:23 UTC+1, nischalshetty wrote: >>> >>> We have been on GAE/J from more than 2 years now. We have 2 products >>> that run on it. A couple of weeks ago, I noticed an unusually high latency >>> for one of our products and as a result a high number of instances being >>> present (I guess if latency increases, the number of instances would >>> increase as well to serve new requests). >>> >>> I logged a production issue >>> (link<http://code.google.com/p/googleappengine/issues/detail?id=7624>) >>> and the gae team took it up swiftly and started work on fixing it. Though >>> it took time to fix it, I was happy that they were in touch while fixing >>> the issue. >>> >>> Since the bump in instances was a result of the problem encountered due >>> to a degradation of GAE infrastructure, I thought it was right on my part >>> to ask for a refund of the extra billing charges that were levied. >>> >>> Our charges are usually in the range of $30 per day but during the 3 >>> days the charges were *$86, $188 and $47* (attached the screenshot). >>> That's pretty steep and it does hurt our weekly budgets as we're a >>> bootstrapped startup. >>> >>> When I contacted customer service and asked for a refund I was told that >>> the SLA is violated when there are exceptions thrown with error code 500. >>> Since that wasn't really the case here, we were denied the refund. >>> >>> In our case it was the latency(caused due to some problem with >>> appengine) that made our app take a big hit which means it isn't covered >>> under SLA! In case the GAE infrastructure degrades again, and instances >>> spin up at a crazy rate once more, it means we have to pay the charges. I >>> dread if this problem ever crops up again and stays for a week. >>> >>> This can happen to anyone due to any bug in appengine and I thought it >>> was good to give a heads up. If GAE causes a high number of instances to >>> spin up for no fault of yours, you would still end up paying the charges. >>> >> -- You received this message because you are subscribed to the Google Groups "Google App Engine" group. To view this discussion on the web visit https://groups.google.com/d/msg/google-appengine/-/Ey_OpgtuhecJ. To post to this group, send email to [email protected]. To unsubscribe from this group, send email to [email protected]. For more options, visit this group at http://groups.google.com/group/google-appengine?hl=en.
