|
Someone know if this bug affects 1.8.6?
|
|
given that there are no known changes in that area, it most likely does
|
|
It does. See https://support.zabbix.com/browse/ZBX-3988?focusedCommentId=45879&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-45879 .
|
|
Still affects 1.8.10. On our system, it crashes several times a day in Firefox 10.0.1.
|
|
Indeed, running 1.8.10 here and it crashes quite often. Running the slideshow directly on a Sony Bravia tv which uses Opera (can't find the exact version number).
We're using 3 screens within the slide:
screen 1: 3xtrigger status + 3xgraph
screen 2: systemstatus
screen 3: URL
I would love to see this getting fixed. Right now we need to manually reload the page whenever the out-of-mem error occurs.
|
|
On our "wall display machine" (Suse 12.2 Laptop), Firefox does not survive the night, Konquerer survives for 2-3 days...
If memory is really free upon a page reload, a fix would be as easy as just redirecting (HTTP 302) to the first slideshow slide after the last slide has been shown...
|
|
Yes that will probably work.
We have implemented the following work-around:
@include/page_header.php in the HTML part:
<meta http-equiv="refresh" content="540">
Quite dirty, but for us it has been working great 
|
|
Yeah, that's a bit crude
We put a page on the webserver
<html>
<head>
</head>
<body>
<p>Page refresh in 1sec ....</p>
<script type="text/javascript">
setTimeout(function(){top.location.replace(top.location.pathname);},1000);
</script>
<!-- note the "top.location" not "window.location". needed to break out of iframe! -->
</body>
</html>
and create a "last slide of every show" with the only content of type "URL", pointing to this page, and added to the end of every slideshow...
Actually, the fix for Zabbix itself could look quite similar ("if last slide ...")
|
|
This "fix" with the refresh page seems to work for me.
i updated from 2.0.2 to 2.0.6 when i got almost the same problem.
|
|
It works, it flushed the memory of the client browser. But it loses the full-screen parameter.I have modifieded it to this:
|
|
also reproducible in trunk r38126 with opera 12.16
|
|
Fixed in development branch svn://svn.zabbix.com/branches/dev/ZBX-3878
|
|
(1) I have 10 slides with 1-3 graphs
Test browser - IE8
After one circle + 20 - 464 kb of memory usage
Problem exist not only in Monitoring->Screens->Slide shows
If slides without graphs memory usage is stable
<richlv> i'm also seeing increased memory usage with the dev branch. while smaller than before, here's and example of opera memory usage steadily increasing over time. 
Eduards RESOLVED r.38634
oleg.egorov
Again in IE8, slides refresh time = 1
Test begin in 11:44
Memory usage: 42763
Test end in 11:46
Memory usage: 105 636
And JS ERROR
Webpage error details
User Agent: Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; .NET CLR 1.1.4322; .NET CLR 2.0.50727; .NET CLR 3.0.4506.2152; .NET CLR 3.5.30729)
Timestamp: Fri, 20 Sep 2013 08:47:12 UTC
Message: 'objDims' is null or not an object
Line: 2887
Char: 4
Code: 0
URI: http:
Eduards RESOLVED r.38664,38676
oleg.egorov CLOSED
|
|
Looking forward for this fix, as the front-end seems to leak a lot on our monitoring wall using Chromium browser..
|
|
nope, still not good. in this graph, left hand side shows a slideshow being open, right hand side - inventory page (same browser, same zabbix frontend)
|
|
Fixed in versions pre-2.1.4 (trunk) r.38681, pre-2.0.9rc1 (alpha) r.38683
<richlv> this does not match "fix versions" in the issue - "Fix Version/s: 2.0.9rc1, 2.1.5"
|
|
(2) this resulted in a regression : the blocks in the dashboard can not be rearranged anymore. drag a block once, second attempt fails
vitalij_zab it intersects with task ZBX-6911, I fixed this there, and it will be released after testing.
jelisejev This has been fixed for trunk in ZBX-6911, but we need a separate fix for 2.0.
iivs RESOLVED for 2.0 in svn://svn.zabbix.com/branches/dev/ZBX-3878 r38985
jelisejev CLOSED.
|
|
(3) and at least with opera this might have resulted in an insane disk cache usage - ~ 2gb for zabbix alone. would be nice to investigate and find out whether we can make it not count each image as worth caching separately
|
|
trunk r38866.
as discussed, there's still some memleak :
|
|
then why reporting it as fixed in 2.0.9 release notes? http://www.zabbix.com/rn2.0.9.php
|
|
partial fix was committed for 2.0.9 (that is, memleak should be smaller, but it's still not gone)
|
|
In 2.0.2 it worked all fine for me.
2.0.6 screwed it all up.
what is changed between those versions, isnt the fix just in the old version already?
the partial fix was no fix at all here. systems run out of memory as fast as before.
|
|
on 2.2.1 issue came back
|
|
I confirm that the issue is still there in 2.4.2. It urgently needs a fix because it makes unattended display of slideshows impossible.
|
|
We "solved" the problem on our unattended displays with Igel Thin Clients thatreboot every 4 hours... If we don't do this, the Igel will freeze after about 5 hours.
|
|
Just a comment regarding a potential direction to dig into. According to http://learn.jquery.com/using-jquery-core/data-methods/ :
"There's often data about an element you want to store with the element. In plain JavaScript, you might do this by adding a property to the DOM element, but you'd have to deal with memory leaks in some browsers."
Could it be that we are storing some JavaScript data in DOM elements?
|
|
I think the cause of this bug are the constants pages that are added as source and not cleaned.
|
Generated at Thu Apr 25 13:06:05 EEST 2024 using Jira 9.12.4#9120004-sha1:625303b708afdb767e17cb2838290c41888e9ff0.
memleak.png