1
1
1qaz2wsx3edc2015-03-01 01:39:42
1C-Enterprise
1qaz2wsx3edc, 2015-03-01 01:39:42

How to isolate performance problem 1s 8.2 in RemoteApp mode (win2012r2)?

Good afternoon.
We have the following:
In the Data Center at a remote site (site1 - Europe) in a hyper-v cluster, in addition to the main services (ad, dns, dhcp), a terminal farm based on RDS2012 is deployed. For the pilot, the configuration is (web + broker) + rdsh1 + rdsh2 + license (4-separate vm). For this installation, single sign-on (sso), delivery of remoteapp to win7 clients, connection of printers via GP with filtering by groups, dynamic formation of a list of bases on the 1c platform depending on the account membership in the ad group is configured. Users run remoteapp applications through the start menu, via webmord or go to RR dns record (ip rdsh1,ip rdsh2) in full screen mode (mstsc). During the pilot operation, a lot of problems got out:
1) The most important thing is that after a week of connecting users to the site2 site (Peter), the connected users suddenly lost performance on the 1c8.2 platform (both in Remoteapp mode and in full screen mode, no difference) - freezes when navigating through the menu, performing batch processing , calling the print dialog (up to 10 sec). At the same time, users in site 3 (Moscow) worked fine. There were no complaints from site 2 users about other remoteapp applications. At the same time, there are no errors in the farm logs (rdsh, broker), the load is 10% -15% at most. I interviewed other specialists and found out that there were no problems on either the hyper-v cluster or the 1c server + subd stack. It is also important to note that these users had no problems with 1s on standalone terminal win2008r2 servers (also in Europe), i.e. they simply rolled back to 2008 and continued to work quietly.
Question - how to isolate the problem?
2) A completely random printing error (that is, when sending a document for printing (to a specific printer), the red program draws a window for us with the inscription "print error". In this case, the user can print from another application to this printer. At the same time, another user on this in the same node, in the same database, it can print a document to the printer of the first user.The
question is how to isolate the problem?Our 1c specialists answered that 1c does not log this error.Next, I
will simply describe my negative experience with RDS2012 (I'm not looking for answers).
a) Problem with profiles. For a short time I was happy with the normal operation of the User Profile Disk (there was no problem for 3 months of operation on the home cluster), as one fine day ... that's right - some users (absolutely randomly) began to receive a temp profile on the node, while vhdx (which is important ) was not blocked. It is curious that this problem could either resolve itself (logged out, logged in - everything is OK), or not resolved at all (an association of SID with a temp profile was registered in the registry branch and the user received it every time.
https://social.technet.microsoft .com/Forums/en-US/...
b) The problem of losing control of the application, in other words, the loss of focus. According to my observations, it arose during active work in parallel running seamless windows (1c8, different databases, for example).
blogs.technet.com/b/ai/archive/2012/01/16/remoteap...
forum.infostart.ru/forum16/topic54809
https://social.technet.microsoft.com/Forums/window...

Answer the question

In order to leave comments, you need to log in

2 answer(s)
S
Sergey Kovalev, 2015-06-11
@1qaz2wsx3edc

Get rid of all old Canon printers and any HP LaserJet 11xx series. Their built-in wonderful print control managers simply mock the print manager on the terminal server. Neither driver isolation nor enabling Easy Print help.
The problem with profiles can be caused by the fact that the user profile was deleted without deleting the registry entry. After that, the user will always get a problem with a temporary profile when they sign in.
It is treated by deleting registry keys in the
HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\ProfileList branch
, we catch and demolish all SID keys ending in .bak

1
1qaz2wsx3edc, 2015-03-19
@1qaz2wsx3edc

It was possible to track #2 - "Printing Error". When connecting printers via GP (and it doesn't matter how - through administrative templates or via GPP), after some time 1C "loses" the printer (s), removal and reconnection helps. I did not look at the logs in more detail.
upd.
The problem of performance loss went away by itself, it was not possible to trace who exactly was the culprit.
After about half a year of operation, it can be noted that since 2012R2 there are problems only when using 1c8 + printing. I tried to distribute - office, browsers, 1s8, guarantor, consultant, all sorts of self-written web software
Problem with 1s8 now.
When you call the print window (ctrl-p), the system thinks for 19 seconds. On 2008R2 - instantly opens.
Printers on 2012R2 come according to policies (with filtering by group), in 2008R2 they are installed manually.

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question