| T.R | Title | User | Personal
Name | Date | Lines |
|---|
| 3177.1 | | ACISS2::LENNIG | Dave (N8JCX), MIG, @CYO | Tue Mar 11 1997 17:27 | 18 |
| Are the 40000 entries all on one source node, or scattered around the
network? If all on one node, given there are by default only two DDS
talker processes on a node, attempting to create 8 WSnodes at the same
time is implicitly creating a bottleneck. Generally speaking, you
should only do one WSnode at a time (and use the /copyall then /world
sequence to prevent non-WSnodes from sending search requests to the
partially populated WSnodes). You don't indicate what speed links
exist between the nodes, but assuming WAN links, you are probably
looking at a few weeks to do 8 WSnodes. I tend to recommend
starting the population of a large WSnode on a Friday evening, and
would usually expect it to be complete by Monday morning or so...
The Messaging Practice (used to?) have a tool called the Rapid Node
Replicator, but I don't know if it's been maintained or tested against
the current version of MR, and besides it also had a constraint that
all the DDS objects had to be owned by the DDS Master node as I recall.
Dave
|
| 3177.2 | | ACISS2::LENNIG | Dave (N8JCX), MIG, @CYO | Tue Mar 11 1997 17:30 | 5 |
| Oh yeah, re: tracking the progress.
Watch DDS counters on both source and destination nodes...
Dave
|
| 3177.3 | What els to check ? | EVTAI1::GODARD | | Thu Mar 13 1997 17:11 | 12 |
| Thanks for your response.
Now, the customer wants a DDS specialist on site to track the
progress...
Is there any other things to verify than counters in order to check the
update progress ?
Regards
Monique
|
| 3177.4 | | ACISS2::LENNIG | Dave (N8JCX), MIG, @CYO | Fri Mar 14 1997 21:03 | 5 |
| Well, the size of the outgoing node's network queue file will shrink
over time (when DDS$TIDY runs), and the size of the destination node's
perm_db and perm_aif file will grow over time.
Dave
|
| 3177.5 | unable to add subscriber: is it normal? | EVTAI1::MORVAN | | Wed Mar 26 1997 12:01 | 15 |
| Hi,
I have been on site yesturday. The update is still going on.
The customer has noticed the following pb on each ws that are updating:
he cannot add or modify a local subscriber. No respons when he types a
add built command.
Is this due to the current massive update that locks the inqueue file
or to a pb with this (inqueue) file?
Thanks for your help.
Regards.
Genevieve MORVAN
|
| 3177.6 | | ACISS2::LENNIG | Dave (N8JCX), MIG, @CYO | Wed Mar 26 1997 14:59 | 11 |
| The Input Queue is just that - a queue.
If there are a lot of updates in the queue, then when they try to
do an update, their request get's tacked on the end of the queue
and has to wait in line to be processed.
I'd suggest you investigate where all these updates are coming from.
With the right set of events enabled, you should be able to examine
the dds$log file to see what node they are coming from.
Dave
|