Welcome to the new Gigaspaces XAP forum. To recover your account, please follow these instructions.

Ask Your Question
0

Updated object missing from the space

Hi,
We have the following scenario: From an API we get some "data" updates on a thread, those updates are for a set of specific IDs. We want to push those to clients and for that we make a POJO from those "data" and update them in space based on the ID and also we write a notification object in the space for each registered client. The clients want to get always the last image of the "data" regardless how many updates were since the last time it processed it, so we consume the notifications with a polling container and try to read the "data" at that point based on the ID. The "data" and the notifications are written from the same transaction. With no apparent reason, even with a single client, at some point it just fails to read the "data" from the space: it returns null, then it may be ok again and so on.
Those updates may come with a frequency of milliseconds. I also tryed to see what happens if I attempt to read the "data" with dirty_read modifier but neither that was working. Sometimes it starts directly bugged then may recover, sometimes it starts ok but in time fails. If fails for subsets of IDs, other working ok. There's no fast switching between ok and fail, it is minutes before it stops working and starts working again. It behaves so even with a single client subscription. Adding client subscriptons (resulting in more notifications written at once) looks like a factor that influences the bug to appear, adding another one may make it ok again...but this is not a condition, maybe just a coincidence, as I sayed it happens even with a single client. There wasn't a single notification object missing from processing, just the "data" ones are bugged.
Any idea on what can be the cause of this?
Thanks

This thread was imported from the previous forum.
For your reference, the original is available here

asked 2013-03-08 09:41:18 -0500

lukeh gravatar image

updated 2013-08-08 09:52:00 -0500

jaissefsfex gravatar image
edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
0

Are you using a notify container or polling container ?
Are you running in FIFO mode ?
Do you perform the updates using a transaction?
Are you using write or writeMultiple when updating the objects?
Which XAP release are you using?
Is the space persistent?

It will be great having a test case reproducing the problem.

Shay

answered 2013-03-08 10:46:42 -0500

shay hassidim gravatar image
edit flag offensive delete link more

Comments

Are you using a notify container or polling container ? - polling container Are you running in FIFO mode ? - tha data is FifoSupport.OPERATION, the notifications are not (on any notifications we need the last data image) Do you perform the updates using a transaction? - yes, same transaction is used when updating the data and when writing the notifications Are you using write or writeMultiple when updating the objects? - write, no writeMultiple Which XAP release are you using? - gigaspaces-xap-premium-9.0.0-ga-b6500, sorry, should have mentioned this in the first post Is the space persistent? - no It will be great having a test case reproducing the problem. - we are a bit busy atm and went for a workaround by updating the data in a hash map per id, we still look it up in the space first to log the issue but then we get it form there. Will come back to this and also try to make a simple system to reproduce it when our time permits.

One other thing I tried to see what happens is to add a session listener for NotifyActionType.NOTIFY_ALL on the "data" which just logs the pojo and the action type. When the issue is present it doesn't get notified, nothing is logged. We can see an initial write, updates when it works ok, silence when the issue is taking place, no takes/removes. The issue can happen from start meaning no initial write either in this case.

Edited by: Lucian Hada on Mar 11, 2013 3:02 AM

Edited by: Lucian Hada on Mar 11, 2013 3:03 AM

lukeh gravatar imagelukeh ( 2013-03-11 02:57:29 -0500 )edit

I suggest you approach the support team to investigate this. Something is wrong. This should not happen. Shay

shay hassidim gravatar imageshay hassidim ( 2013-03-13 08:47:58 -0500 )edit

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower

Stats

Asked: 2013-03-08 09:41:18 -0500

Seen: 42 times

Last updated: Mar 08 '13