Summary | Trean extremely slow; probably due to amount of data transferred |
Queue | Trean |
Type | Bug |
State | Resolved |
Priority | 2. Medium |
Owners | ben (at) |
Requester | wk (at) mailstation (dot) de |
Created | 05/06/2005 (7383 days ago) |
Due | |
Updated | 01/21/2006 (7123 days ago) |
Assigned | 01/21/2006 (7123 days ago) |
Resolved | 01/21/2006 (7123 days ago) |
Github Issue Link | |
Github Pull Request | |
Milestone | |
Patch | No |
but we're now talking about optimizing things and not making Trean
usable. :-)
Thank you very much. This is greatly appreciated. From my point of
view this issue is resolved.
State ⇒ Feedback
completely. See the DataTree browser for an example.
problem with trean. That many categories is just going to produce a
lot of html/javascript code. I don't have any clever ideas on how to
fix this. Suggestions?
Anyway, there are about 280 categories. Maximum depth is 10.
tree on the left of the page in trean? That's the part that's
generating the large html files.
bookmarks don't seem to have this problem.
As for the categories: I'm not sure. I can run any SQL statement you
want to verify the precise numbers, though. (I'm afraid I still don't
fully understand the Horde datatree.)
State ⇒ Feedback
that you have 1600 bookmarks as a user, or total among all users?
Also, how many categories are being viewed?
Priority ⇒ 2. Medium
uncollapse a collapsed part of the tree after changing the renderer.
Anyway, the problem is still the same.
the whole category tree by default.
slowly, of course.
browse.php, line 42.
the attached logs, too.
State ⇒ Feedback
really dig into this, yet.
Trean in its current state is simply unusable with larger bookmark
collections.
My users think about starting a revolt. ;-)
Seriously, I need to make a decision about temporarily replacing Trean
with something else or appeasing people. I've no problem waiting for a
fix but I need to know *if* there will be one sooner or later.
New Attachment: trean_http_log.zip
realize that even if only part of the category tree is being *shown*,
it's being completely rebuilt including all its leafs, those not
shown, too, on virtually any action.
On entering Trean: Would it be possible to build only those parts of
the tree that are shown? I realize that would mean you would have to
load other parts dynamically when the user uncollapses new leafs.
On entering a category: I don't really see why the whole category tree
would have to be rebuilt. What am I missing here?
I've attached a ZIP archive containing two logs of the http streams.
One for entering Trean, the other for opening a category.
Unfortunately, it didn't help:
Startting Trean: "GET /horde/trean/ HTTP/1.1" 200 241005
Entering a category with only one bookmark: "GET
/horde/trean/browse.php?category_id=1810 HTTP/1.1" 200 236853
So getFavicon doesn't seem to be the culprit. I'm now going to
temporarily switch from https to http and sniff out what exactly is
being transferred over the network. That should hopefully help finding
the problem.
In trean/lib/Trean.php, getFavicon() function, line 242, add a line
containing "return '';"
I noticed that this function is a bit of a bottleneck while I was
working on the search feature. Let me know what the results are.
Priority ⇒ 3. High
State ⇒ Unconfirmed
Queue ⇒ Trean
Summary ⇒ Trean extremely slow; probably due to amount of data transferred
Type ⇒ Bug
the following problem *seems* to have started with the option to start
Trean with a collapsed category tree.
Opening the the "Browse" view, all categories completely collapsed,
takes about 80 seconds. There are 1617 bookmarks in the database.
It takes as long to enter any category.
Checking my Apache logs for errors I didn't find any but the amount of
data transferred when opening Trean or any category is so immense that
I can only consider it a bug:
1. Opening Trean:
"GET /horde/trean/browse.php HTTP/1.1" 200 238033
2. Entering a category:
"GET /horde/trean/browse.php?category_id=1963 HTTP/1.1" 200 238149
3. Switching directly to another category:
"GET /horde/trean/browse.php?category_id=1780 HTTP/1.1" 200 238628
More than 230kb per request doesn't really sound right.