cancel
Showing results for 
Search instead for 
Did you mean: 
Create Post
Level 12

V9 Upgrades - Let's here the stories

Now that V9 is out I wanted to start a thread on how upgrades from 8.5.X to 9 have been going?  I'm especially interested in people running multiple pollers.

I'm going to wait a could weeks before I upgrade so the bleeding edge folks can work out any issues. 😉

Good luck.

BB

0 Kudos
161 Replies
Level 11

I just upgraded to newest 9.5 version and I just want to say finally there is an icon for SSH on the integrationicons menu. Everytime I would upgrade I would have to

 reconfigure for SSH however we use SSH version 2 and the link for SSH is ssh:\\ For version 2 it must be ssh2:\\ so I had to change the integrationicon file but

at least all I had to do was add the 2 after ssh and now it works great. Just thought I would put my 2 cents worth in.

0 Kudos
Level 10

I just upgraded to 9.  It went very well, except for two things.

I had to do a repair on APM to get it working, and then I had to repair the NPM install to get UnDP working.

Other than that, quite smooth and very fast.  No more 15 minute wait while configuring the database.

Justin
 

0 Kudos
Level 9

 Now that is seems there are quite a few fixes out are they all available through the update included in V9? We are looking at moving off of 8.5.1 in the next few weeks and was hoping I wouldn't have to find a dozen different patches to fix graphs, etc.

0 Kudos
Level 13

Bryan, probably a little late for your upgrade - however we didn't have any issues with our alerts or maps. As long as you have a database backup, and copy of the Maps/Reports folder I think you will be OK.

I shutdown all the pollers and upgraded the primary SLX server first - then did the pollers. Just had the normal issue of the install starting all the services - and having to shut them down again so I could upgrade the pollers (I wish there was an install option to leave the services down).

We made the choice a long time ago to run the database in simple mode (i.e. no transaction logs). We found this took the pressure off the database server - but obviously has a recovery trade-off (i.e. we can only recover to the last backup if required).

Dave. 

0 Kudos
Level 12

 Thanks Dave....I did what you suggested.  Here are some thoughts :

  • Upgrading multiple servers is a huge pain, constantly having to upgrade one and stop the services again so I can go to the 2nd box
  • I was very surprised that the database part was really quick...I have 12K elements and it was done in like 30 seconds
  • Got some weird errors on the site but I decided to reboot all my servers 1 last time and things look better

Support items

  • Universal device poller will not load...period.  I get the pop-up box and it just sits there.  All of my Custom MIB pollers do not work at this point.
  • Clicking on a node takes a VERY long time to load the device details page
I'm just glad I have a working web site at this point.  I'll work with support on these items and see what happens.

BB
0 Kudos
Level 13

Glad to hear your upgrade went OK.


I don't think you waited long enough for the Universal Device Poller to load - ours takes anywhere from 13-15 minutes with 15K elements - which is a very similar size to yourself (and 174meg of memory once loaded) - development is working on this for the next service pack.


Node details do take longer than 8.5.1 - not quite sure why, but it's not something I have reported as I don't have any hard numbers (and I didn't see anyone else making a similar comment).


Dave.

0 Kudos
Level 12

I opened a ticket for the UnDP issue and loading of the node details.  I launched UnDP on one of my pollers and after an hour it still isn't loaded.  Loading of node details takes between 70-80 seconds on my system.  Everything else (like interfaces, etc) load in 1-2 seconds.

Node management on the web site is nice to have although I find it pretty slow to work with.  It could just be my system so I'll have to have support gotomeeting and see what's going on.

BB

0 Kudos
Level 13

70 or 80 seconds - wow, we aren't in the same ballpark. When I said ours were slower, I estimate that typically the node details screen took 2-3 seconds under 8.5.1, and 5-6 seconds under 9.0 (but I can't back the old numbers up with hard data). Any of the other screens (i.e. interfaces) seem to be unchanged in performance terms from the previous release.


For the UnDP poller loading, does the process show memory growth during this hour? Process memory growth may actually show that it's still loading and not stalled or hung.


I'm be interested to see the result of both issues.


On another v9 observation...


The way MapMaker now expands the list of devices (once a vendor is selected) is very annoying. For a large list of Cisco devices it can take several minutes for the list to populate. I can see that for some users this may make the inital load faster - but it's bad news for the larger installations.


Dave.

0 Kudos
Level 12



70 or 80 seconds - wow, we aren't in the same ballpark. When I said ours were slower, I estimate that typically the node details screen took 2-3 seconds under 8.5.1, and 5-6 seconds under 9.0 (but I can't back the old numbers up with hard data). Any of the other screens (i.e. interfaces) seem to be unchanged in performance terms from the previous release.

For the UnDP poller loading, does the process show memory growth during this hour? Process memory growth may actually show that it's still loading and not stalled or hung.

I'm be interested to see the result of both issues.

On another v9 observation...

The way MapMaker now expands the list of devices (once a vendor is selected) is very annoying. For a large list of Cisco devices it can take several minutes for the list to populate. I can see that for some users this may make the inital load faster - but it's bad news for the larger installations.

Dave.

 

Web Site - only node details seems to be affected.  Interfaces, etc load up in 1-3 seconds.  We did find a large amount of "white" space in the DB leftover from a previous running of NetFlow.  The DB was up to 140GB with only 7GB actually used.  We ran a shrink and got it down to 37GB.

UnDP - Yes I do see the memory usage grow as it's loading but still takes a very long time to load.

MapMaker - I think this was changed in V8.5...I hate it too.

BB

0 Kudos
Level 12

I think we have identified the issue...fixing it I'm not sure.  It seems that after I removed the custom mib gauges/charts for node details the node details paged loads in like 3-5 seconds.  I'm not sure of the conflict from custom mib poller to UnDP but something isn't right.  I sent support the files so hopefully something can turn up.  UnDP does load in about 2 minutes now which seems to be faster than before.

BB

0 Kudos
Level 15

I upgraded my SL2000 server last week.
The upgrade took 30 minutes and went without a hitch.
The website has been running fine since the upgrade.


The issue I have is the links to the charts no longer work in report scheduler.
I am e-mailing out charts of connected VPN users, but the new asp.net URL format no longer allows me to retrieve the chart with Report Scheduler and mail it out weekly:

Previous to the upgrade the URL looked like this:
http://orion-sl2000/NetPerfMon/ViewChart.asp?Chart=CustomOIDChart&NetObject=N:2515&CustomPollerID={FBB9360E-A013-4EEF-8A9A-5EFB690BC559}&SubsetColor=FF0000&RYSubsetColor=0&Period=Last%207%20Days&SampleSize=15M&Title=VPN+users+%2D+last+7+days&ReBuild=TRUE&FontSize=Medium&Width=640&Height=0 

The URL now looks like this:
http://orion-sl2000/Orion/Controls/ChartImagePipe.aspx?ChartID=customChart$ctl01$chart_Base&KxRx=0x09556
but no chart is displayed.


This is a show-stopper for my SLX upgrade.
If I can't e-mail charts, there is no way I can upgrade to V9.

0 Kudos
Level 19

This will be fixed in the next service pack. There will be a new URL for charts that (like the old one) supplies all of the parameters needed to generate the chart instead of using ChartImagePipe.aspx and relying on the asp.net session.

0 Kudos
Level 7

I have no seen this issue any where else, so I think I'm just doing something wrong.  We got a new server for Orion so I decided to just install the new version 9.  The install went fairly smoothly.  The only issue I'm really seeing now is that I had a number of reports that I created and had published to the website.  None of these reports show up any longer.  (Yes, I moved them from the old server! :))  If I open report writer on the server all of the reports (canned and custom) are there.  If I open a custom report that doesn't show up on the web site everything looks as it should.  The box is checked that says to publish to the web.  I've stopped and started resources on the server and I've even rebooted the server and I cannot get the reports to show up on the web.

 What am I doing wrong?

0 Kudos
Level 19

Sarah,

It could be that the solarwinds_website user lacks permission to see the report files you copied. To check this, open the file properties in Explorer and look at the security tab. Compare the settings for a file that does show up on the website with the settings for one that doesn't.

0 Kudos
Level 7

Sarah,

It could be that the solarwinds_website user lacks permission to see the report files you copied. To check this, open the file properties in Explorer and look at the security tab. Compare the settings for a file that does show up on the website with the settings for one that doesn't.



Tim - I've been in contact with SW support.  I've checked permissions for the solarwinds_website user as well as the iusr_(machine name) user.  I have also re-run the website configuration wizard to recreate the COM+ object.  I've cleared website cache and restarted IIS on the webserver.  I'm working to schedule a GoTo meeting for hopefully tomorrow.  I'll post what I find out.  It looks like I'm having some issues with alerting as well.  Thanks!!  🙂

0 Kudos
Level 7

Okay, so here's the deal on my reports not working.  I finally opened a ticket with support.  They said that this happens sometimes during upgrades, but it's fairly rare.  I moved the database from one server to another and he said that that may have had something to do with it as well.  Anyway, I was able to copy the reports out of the reports folder and paste them on to my desktop and then delete the contents of the reports folder and then copy everything back in again.  Then they all published to the web.  Simple, yet so complicated.


 On another note, I originally had my alerts temporarily disabled and I figured that was the reason that I was not receiving alerts (and I'm sure it had something to do with it). But now that I've fixed that, still.. none of my alerts work.  I was hoping it would be a simple matter of recreating the alerts to fix this problem but not the case.  I've recreated 2 alerts from scratch.  If I use the test alert option, the test sends just fine but if I literally unplug something out of the network that I'm alerting on, I get nothing.  This has been the single most frustrating upgrade for this software.  I've never had issues before. 

0 Kudos
Level 10

I had the same problem with alerts. In my case the solution was: Open NPM, click ALERTS button, remove checkmark from box labeled, "Temporarily Disable alll actions for ALL Basic Alerts". How it got checked in the first place is beyond me.

0 Kudos
Level 7

I had the same problem with alerts. In my case the solution was: Open NPM, click ALERTS button, remove checkmark from box labeled, "Temporarily Disable alll actions for ALL Basic Alerts". How it got checked in the first place is beyond me.



Yes, same here, but even after I removed that check they still don't work.

0 Kudos
Level 17

spandey- 

Did you do a migration and then updgrade?

0 Kudos
Level 7

spandey- 

Did you do a migration and then updgrade?



Migrate and upgrade? Probably not.  I got a new server for the SW App.  Since the new version was availble for download, I downloaded version9.  The server team wanted to also move the database so they moved it to the same server.  Then I installed the software, attached to the db, etc. 

0 Kudos