captaingadget
IS-IT--Management
Am I correct in saying that routine CPU switch-overs no longer occur with release 4.5 software?
Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations derfloh on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
OP 81 CPU Switch-over 2
- Thread starter captaingadget
- Start date
- Status
-
1
- #2
bigindian65
Technical User
It no longer occurs once you go to Pentium proccessors. Can be on 3.0
-
1
- #3
Telephonist
Technical User
This is incorrect.
In LD 17, in the OVLY Gate Opener, there is a prompt as follows:
- MID_SCPU (NO) YES Midnight Switch Cores
Allow or deny Midnight Switch Cores, where:
Deny causes the system to perform the 3PE test during the
Midnight routine instead of switching CPUs.
Allow causes the system to switch CPUs during the
Midnight routine instead of performing the 3PE test.
Note: Applicable to CPP systems only.
The default is to run the 3PE test and not switch cores, but changing the prompt to YES allows the system to switch cores during the midnights.
In LD 17, in the OVLY Gate Opener, there is a prompt as follows:
- MID_SCPU (NO) YES Midnight Switch Cores
Allow or deny Midnight Switch Cores, where:
Deny causes the system to perform the 3PE test during the
Midnight routine instead of switching CPUs.
Allow causes the system to switch CPUs during the
Midnight routine instead of performing the 3PE test.
Note: Applicable to CPP systems only.
The default is to run the 3PE test and not switch cores, but changing the prompt to YES allows the system to switch cores during the midnights.
- Thread starter
- #4
bigindian65
Technical User
Hey phonist, it IS correct because once on pentium Nortel does not want cpu switching. I didn't say it could not be done.
Nortel definitely does not want you switching cores on the pentium processors during midnights. Just got off the phone with a Nortel tech yesterday about this as we had an issue with a bad CPP card. I asked the question directly and he replied pretty sternly NO. According to him it is not necessary due to the health monitoring.
One thing to note is you should do a switch once in a while as a just in case measure. We found our ethernet port on Core 1 was bad and when we had a failure of Core 0 our Call Pilot couldn't communicate with Core 1.
One thing to note is you should do a switch once in a while as a just in case measure. We found our ethernet port on Core 1 was bad and when we had a failure of Core 0 our Call Pilot couldn't communicate with Core 1.
to cut to the chase, only switch during a trouble clearing mode.. but pay very close attention to the health of both the active and the standby side.. i perfer to run on 2 cpu's, switching during midn.. but i would not push a pentium into that mode...
one of the 1st things ive' always done is to make sure system trouble either is or isn't side related...
john poole
bellsouth business
columbia,sc
one of the 1st things ive' always done is to make sure system trouble either is or isn't side related...
john poole
bellsouth business
columbia,sc
Telephonist
Technical User
I asked the question directly and he replied pretty sternly NO."
Did the Nortel tech say why? If they don't want you to switch, there has to be a reason, other than "it's not necessary". What is the potential issue? We're switching cores nightly on our CPP4 processors, so I'd like to know why Nortel doesn't want it done, particularly since it can be enabled so easily.
If the ethernet port on a core can be bad, and there is no indication in the health monitoring, that can be a pretty big uh-oh when a hardware failure forces a core switch-over, especially since everything is getting connected via ethernet these days.
Did the Nortel tech say why? If they don't want you to switch, there has to be a reason, other than "it's not necessary". What is the potential issue? We're switching cores nightly on our CPP4 processors, so I'd like to know why Nortel doesn't want it done, particularly since it can be enabled so easily.
If the ethernet port on a core can be bad, and there is no indication in the health monitoring, that can be a pretty big uh-oh when a hardware failure forces a core switch-over, especially since everything is getting connected via ethernet these days.
bigindian65
Technical User
look, it was re-written for a reason. It's not like Nortel said opps we forgot to code the cpu's to switch on pentium proccessors.
Telephonist
Technical User
OK, so the question still stands - what is the reason?
After our CPP4-FNF upgrade, we had an issue with the SPLIT command. The cores would not split, even though the health state of both cores was a full 36, and they were showing as running in true redundant mode. It turns out that we had a faulty utility card in core 1. The issue was escalated to Nortel by Black Box TAC, and there were at least two Nortel engineers working on it. They knew we were running our midnights with the core swap enabled, and said nothing about it.
BTW, the faulty utility card didn't show up in the health state monitoring.
Johnpoole, you say that you prefer to run with CPU swapping during the mids, but you wouldn't push pentium processors into that mode. Can you say why? - in other words, what do you know that we don't know?
There are a dozen technical people servicing a half-dozen redundant switches in the State network, who are all interested in this topic.
After our CPP4-FNF upgrade, we had an issue with the SPLIT command. The cores would not split, even though the health state of both cores was a full 36, and they were showing as running in true redundant mode. It turns out that we had a faulty utility card in core 1. The issue was escalated to Nortel by Black Box TAC, and there were at least two Nortel engineers working on it. They knew we were running our midnights with the core swap enabled, and said nothing about it.
BTW, the faulty utility card didn't show up in the health state monitoring.
Johnpoole, you say that you prefer to run with CPU swapping during the mids, but you wouldn't push pentium processors into that mode. Can you say why? - in other words, what do you know that we don't know?
There are a dozen technical people servicing a half-dozen redundant switches in the State network, who are all interested in this topic.
bigindian65
Technical User
Telephonist, I talked to a friend of mine who got it explained by Nortel, I'm just not sure I can correctly relay it. Core 0 & 1 motorola - ran the show, if on CPU 0 it had resource control of the switch.
Pentiums, share all informatoin , constantly talking to each other, checking each health tiers, so they share resources if one has a healther stack, without a swap over. It kinda makes sense. That is why there is no needto swap, but if you don't pay attention to the health of each, and particuliarly the Elan/Ethernet performance, then yes, a force swap could hurt you. CPU 0 can be active, and be letting CPU 1 drive a healther resource. I hope that helps a little.
Pentiums, share all informatoin , constantly talking to each other, checking each health tiers, so they share resources if one has a healther stack, without a swap over. It kinda makes sense. That is why there is no needto swap, but if you don't pay attention to the health of each, and particuliarly the Elan/Ethernet performance, then yes, a force swap could hurt you. CPU 0 can be active, and be letting CPU 1 drive a healther resource. I hope that helps a little.
- Status
Similar threads
- Locked
- Question
- Locked
- Question
- Question