We're in the middle of deploying new catalog PCs which will use Aquabrowser (www.aquabrowser.com) for catalog searches. The browser uses a standard search form with a Flash-based "context" tree that comes up on the left side of the screen after you've entered search text. So you can test this, our implementation can be found here:
plymouthlibdist.aquabrowser.com
When we perform the steps below, the browser "hangs up". By hanging up, I don't mean that it crashes - the application is still displayed as "Running". not "Not Responding". You just can't do anything with the application for a bit. After anywhere from 30 seconds to a couple of minutes to indefinitely, the browser begins responding again. Here are the steps:
1. Enter search text - "houses" for example. Items will be displayed and the context tree will appear.
2. Click on the 'Library Homepage' link above the search field. This will take you to the homepage.
3. Click the 'Back' button on the browser toolbar. This takes you back to the catalog. This is where the browser hangs. As mentioned above, Task Manager shows the app still running but all is locked up and remains that way for a while.
We've tested this in IE 6 extensively and can do this all day long without any lockups. We also tested this with the same version of PWB and a vanilla .ini and still had the lockups. Any advice?
PWB 2.10.6 - reproducible "hang-up" problem
Moderators: Tyler, Scott, PWB v2 Moderator
-
Joseph Stude
- Participant
- Posts: 16
- Joined: Wed May 17, 2006 7:28 am
-
Joseph Stude
- Participant
- Posts: 16
- Joined: Wed May 17, 2006 7:28 am
We've isolated this to the PAC hardware. For some strange reason, doing a search for 'houses' works fine on my own work computer but when following the steps I originally posted it locks up every time on the PAC. We've exhausted every possibility we can think of, including making sure Java and Flash are up-to-date, making sure all drivers are current, and even checking the BIOS version. We also tried this with IE7 and then IE8 with no change in results. I need to reiterate - we can make this work all day long on the PAC hardware using IE itself... it only locks up when running through PWB.
Something else interesting to note - doing obscure searches for things like "rumplestiltskin" yields fewer result terms on the "word cloud" on the left than something like 'houses' does. I've found that doing these more obscure searches works literally every time in PWB, but the lockups always occur when the cloud has more result terms in it. The browser uses Flash to render that portion of the page - perhaps there's something wacky with the way PWB is rendering and/or dealing with Flash objects?
Something else interesting to note - doing obscure searches for things like "rumplestiltskin" yields fewer result terms on the "word cloud" on the left than something like 'houses' does. I've found that doing these more obscure searches works literally every time in PWB, but the lockups always occur when the cloud has more result terms in it. The browser uses Flash to render that portion of the page - perhaps there's something wacky with the way PWB is rendering and/or dealing with Flash objects?
-
Joseph Stude
- Participant
- Posts: 16
- Joined: Wed May 17, 2006 7:28 am
-
Joseph Stude
- Participant
- Posts: 16
- Joined: Wed May 17, 2006 7:28 am
Scott,
After much wailing and gnashing of teeth, we've isolated it to problems with our (offsite) aquabrowser site only on this particular hardware profile. Something with the way the site is drawn using a particular Flash object is sending one of the two cpu cores skyrocketing to 100% cpu utilization, where it stays. Thus the browser isn't crashing, but there are no cpu cycles left to give it so it appears to hang for long periods of time w/o accepting input. We can go to other libraries' aquabrowser sites all day long using that same hardware w/o a problem. They're investigating but so far they haven't found anything...
After much wailing and gnashing of teeth, we've isolated it to problems with our (offsite) aquabrowser site only on this particular hardware profile. Something with the way the site is drawn using a particular Flash object is sending one of the two cpu cores skyrocketing to 100% cpu utilization, where it stays. Thus the browser isn't crashing, but there are no cpu cycles left to give it so it appears to hang for long periods of time w/o accepting input. We can go to other libraries' aquabrowser sites all day long using that same hardware w/o a problem. They're investigating but so far they haven't found anything...