Good day,
2 years ago, my application hit OOME issue, I analyze the heap dump and found that its cause by com/ibm/xml/xlxp2/api/wssec/WSSXMLStreamReader, I tried to google around and found this IBM article:
https://www.ibm.com/support/pages/apar/PM42465
Base on what I understand from this article, I need to add in a system property which is call com.ibm.xml.xlxp2.api.util.Pool.STRONG_REFERENCE_POOL_MAXIMUM_SIZE and set its value to 0.
Thus, I set this in Application Servers > myApp > Process Definition > Java Virtual Machine > Custom Properties:
I restart the server and so far didnt see this OOME issue happen again for 2 years. (In fact I am not sure this solution is work or not as I also want to keep in view)
Last week, there is a OOME issue reported again, and I check the heapdump, found that its same issue again.
Looks like add the system property is not work.
May I know who to fix this OOME issue? Is the step I did is correct? My WAS version is 8.5.5.
Kindly advise, appreciate your help.
Related
Hoping to get some insight into the behavior I am seeing while trying to use GameLift Matchmaking.
I have my configuration setup as such that it does not require player acceptance, as such:
GameLiftMatchmakingConfiguration:
Type: AWS::GameLift::MatchmakingConfiguration
Properties:
AcceptanceRequired: false
...
When I go to the GameLift console and into the configuration I see that it was correctly set as well that it does not require acceptance.
This is where I am confused, because now I have it working where it places 2 users in PotentialMatchCreated and I get this event from GameLift. Then 30 seconds later, I get more events stating that these placements timed out and searching again.
The configuration documentation states that AcceptanceTimeoutSeconds is only required if AcceptanceRequired is true, which it is not for me.
the acceptance documentation states that you only call this When FlexMatch builds a match, all the matchmaking tickets involved in the proposed match are placed into status REQUIRES_ACCEPTANCE
Which its not, its in PotentialMatchCreated.
So my question is, what do I have to do to confirm a placement once GameLift places 2 users into a match? I am a bit surprised because I thought that the fact that it doesn't have to be accepted would mean that its automatically accepted match.
Also theres very little documentation I found regarding what to do in this situation, given the nature of this service not being as known as others I totally expected that but really hoping someone can help me on what to do next.
Any insight or help is greatly appreciated.
UPDATE1:
Additional information: I do not need to utilize GameLift fleets or builds at all. We have a browser game we are building and just want to utilize the matchmaking feature. So we dont have any game servers or anything like that, its just on our website where they would play the game and use our api's/websockets that puts the matchmaking on the server and notify the client when a match has been found with all the subsequent details.
UPDATE2:
To confirm my suspicions I decided to actually try to use the accept match endpoint and see what happens. Just as the documentation states, you can only accept a match if it requires acceptance. I get an error stating that I cannot accept a match that is not in REQUIRE_ACCEPTANCE state. Guessing this is a bug on AWS's side, I don't see any other endpoints that I can hit for being in state PotentialMatchCreated.
Figured out the issue. It has to do with the FlexModeMatch on the GameLiftMatchmakingConfiguration. For my use case, just needing matchmaking, STANDALONE is the correct implementation because we aren't having GameLift actually create game servers/sessions for us. I had mine using WITH_QUEUE which is why I believe I was having issues. Seemingly working correctly now.
First question to gain some privileges...
I hope someone can answer it, maybe some links where I could find the info would be great, or the reason why is not a good idea hehe!.
I've tried running it as a service and worked like a charm for around 2 weeks, but then I encountered an issue whose responsible I haven't found.
Nevertheless, this one is a nice guide:
https://naysan.ca/2019/09/07/jupyter-notebook-as-a-service-on-ubuntu-18-04-with-python-3/
Grakn is awesome, but it also seems very unstable.
The exactly same initial import queries -run in another keyspace- (meaning both were empty before) are only working in one keyframe, the other throws this stupid error:
Error status: 400, error info: facebookCategory doesn't have an 'isa', a 'sub' or an 'id'
In my scheme facebookCategory is defined like this facebookCategory sub category. Category is defined as category is-abstract sub entity.
So - what's wrong with Grakn?
This is a known issue we are working on right now:
when you start migrating data on a keyspace and for some reason your schema or data are wrong/missing, the keyspace sometimes gets corrupted. So when you will try to reload data - even if now everything is correct - you will get that error, again because the keyspace is corrupted.
We are working on this issue internally to solve it as soon as possible.
The error you get is trying to tell you that there is a corrupted schema, we will try to improve it and make it more explicit so that you don't think it's stupid.
We know Grakn is still far from perfect but we can assure you we are just trying to fix as many bugs as possible.
Please try to migrate everything in a clean keyspace, you can find also more issues discussed in our github repo and our support forum
THIS is not a duplicate question!
In my app, I have collections directory (in project root) with files (every file corresponds to one collection). My tempates (in client/templates/...), of course, use collections. Everything perfect with one small exception. One of my collections is undefined but the rest is ok. Every collection is defined in the same directory and in the same directory level. Where is the problem?
EDIT: I tryed inserting collections inside lib. Not works. I tryed deep nesting. Not works.
EDIT: Tell me, why I have 2 downvotes. Downvotes are for good for nothing questions, this question may help in future another Meteor users. So, why are you downvoting this quetion?
Bug has interesting roots. I'm using fedora 21, and in theese times, Chrome has bug which causes complete freeze (1-5 hours). This bug has many Linux distributions. After I restarted chrome after freeze, the error vanished. So, the result is that Chrome has two bugs instead of one :)
EDIT: No. It's premature conclusion. It works only for a while but then it failed again. In Chrome and in Firefox too. I'm really confused.
Just wondering if any one might now how to fix this issue.
Every part of the databroker is created and also all the database connectivity is functional.
But once we try to place them together we come up with a error saying cannot create Null Databroker.
The thing that is really strange is that we have databrokers that are basicly the exact same working.
Can any one shine some light on this issue?
We worked on this together today. We did three things to troubleshoot and fix the problem:
we changed the calling workflow so that the dataset is refreshed inside the show-screen tags (previously, this was done before the screen was called).
we refreshed the dataset (a H2 db) and connected the database (before, it showed as pool not open)
we changed the database name to main
It is working now.
"Null databroker" suggests that the broker was not created correctly, although it doesn't look like that was the case since you don't mention having to modify it to get it to work.
In these cases it can be useful to look at the error logs through the admin app to see if there are errors being thrown at some time other than compile time.