| Title: | DECmcc user notes file. Does not replace IPMT. |
| Notice: | Use IPMT for problems. Newsletter location in note 6187 |
| Moderator: | TAEC::BEROUD |
| Created: | Mon Aug 21 1989 |
| Last Modified: | Wed Jun 04 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 6497 |
| Total number of notes: | 27359 |
Hello, A while back PA was used to produce statistics on a circuit for past time. Circuit counters were polled hourly. The PA command used was something like: show node4 xxxx circuit syn-0 all statis, for start 6:0 every 1:0 - until 10:0 dur 1:0, in domain .yyy The command works well if there are no polling problems and the data is stored normally. But if a poll is missed for whatever reason in the scope of interest timeframe PA ACCVIO's. Sometimes you are kicked out of MCC sometimes not. If you are not and you retain the MCC prompt then *any* subsequent PA commands (whether the data is stored properly or not) ACCVIO's again. This continues until MCC is exited which means PA is useless until another MCC session is opened. In the above example the only way to get stats for the time frame is to increase the interval (ie every=2:0 and dur=2:0). Because missed polls appear to happen occasionally with MCC (for different reasons) this should be fixed. PA at a minimum should bow out gracefully, or better yet go to the next poll and produce a stat for the longer duration automatically. If you need an actual example let me know and I'll see if I can recreate it. This error is consistent so you can probably recreate the situation there also. best regards, brad...
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 1449.1 | We got it too! | TOOK::ANWARUDDIN | Anwar | Thu Sep 05 1991 11:18 | 12 |
We stumbled across this problem last week. What we observed was that 1. If the duration is < the poll period, PA ACCVIO'd. 2. If the start time is not specified, i.e, start = 12-Jun-1991, (default start time is midnight) then it worked OK even with duration < than poll period. 3. If the start time was >= the poll period everything was fine. I don't think there were any missed polls in the data we were testing with. This problem will be fixed. | |||||
| 1449.2 | thanks | PATS::WOODCOCK | Thu Sep 05 1991 12:56 | 15 | |
>>> 1. If the duration is < the poll period, PA ACCVIO'd.
Hi Anwar,
A slight word change in the above statement may be a bit more accurate and help
when you go looking for the problem. If a poll is missed this results
in an actual larger poll period in reality. Therefore the problem statement
could read:
1. If the duration is < the "interval of successfully stored data",
PA ACCVIO'd.
thanks for listening, this one hurts for production batch jobs,
brad...
| |||||