TIMEPROMPTWAIT

[Wim Van den Wyngaert]

I have very old AXP500/400.
When they boot after a weekend power off, they prompt for the date. Probably due to bad battery.

If I set TIMEPROMPTWAIT to 0, will it continue without prompting ? Will it take the very-old date (17xx ?) as default ? Even when the battery is bad ?

Wim

Wim

[Ian Miller.]

I think it always waits if the battery is bad.

[Robert Gezelter]

Wim,

According to the help text for TIMEPROMPTWAIT in SYSGEN, the following is the behavior (the text is still the same in 8.2) [paraphrased for berevity]:

== 0 - No prompt or wait occurs; use last book as system time
0 > TIMEPROMPTWAIT > 32768 - One prompt is issued; the system waits TIMEPROMPTWAIT seconds; then proceeds using the last boot time as the time
32768 <= TIMEPROMPTWAIT < 65535 - Time waits for 2 seconds; then doubles until 256 seconds; and then the sequence is repeated indefinitely

WADU, I recommend checking the HELP text directly and not rely on my paraphrase.

- Bob Gezelter, http://www.rlgsc.com

[Wim Van den Wyngaert]

During lunch break, I shut 1 node that had the problem after putting TIMEP on 30. Then power off for 1 minute and reboot. No prompt at all. Tried it 2 times

The strange thing is that none of the stations had it before and now 3 out 4 stations had it.

Could it be related to the fact that the station was power off during the time change ? I use DTSS.

Wim

[Hoff]

TIMEPROMPTWAIT and SETTIME are the two system parameters.

By "AXP500/400", did you intend "AlphaStation 500/400"? (Some of these system model names are unfortunately really close; I've learned to be cautious about system marketing model identification.)

Many of these boxes used a Dallas or similar chip, and the chip contains an internal lithium battery. When these lithium cells fail (and that usually happens somewhere between five and usually closer to ten years), the box forgets what console to use and generally selects AlphaBIOS/ARC and not SRM, forgets the system time, and forgets the environment variable settings.

Field identification for the Dallas chip:

http://www.futurlec.com/Dallas/DS12887.shtml

Or dig up the service manual and go looking.

If you find you have a Dalla chip in this box, it's socketed on every box I've encountered.

Some of the oldest boxes used a NiCd battery pack or a lithium coin cell, and these too could drain. Check the module(s) in the box, or find the service manual and have a look.

Stephen Hoffman
HoffmanLabs

[Hoff]

DTSS doesn't get involved until rather further along in the boot process -- the saved time and such -- and there are some sanity checks -- involves the BB_WATCH chip and the saved system image time. (And AFAIK, both DTSS and NTP distribute UTC. Not local time.)

I haven't looked at the delta for declaring the saved time value and the BB_WATCH value too "extreme" in some time, but it's conceivable that the year might have changed in the saved system image. We're also in the window when the saved values get rewritten.

(VAX systems are notorious for various operational burbles within the first several months of the year, given the 497 day maximum resolution of the VAX TOY. Alpha BB_WATCH handling is rather better than the VAX TOY handling, however. The FAQ has details.)

But if you can't duplicate it, it may well be a stale rechargeable battery that got recharged, or a weak battery that got warmed up. Close it out as a PotM dependency (phase of the moon), err, as not being reproducible?

[Wim Van den Wyngaert]

Yes it's an AS500/400. I most of the time prefer AXP because Alpha means development in the bank.

I'll be in holiday next week and power off my station to see how it reacts.

Didn't know that the battery was recharable.
And yes, they are over 10 years old.

Wim

[Wim Van den Wyngaert]

I'm also aware that bad battery causes console variable loss. May be it's not the same battery because this was not the case. And the really bad battery problem occurs very rarely, so it seems almost impossible that it is the battery because 3 out 4 stations had the problem.
So, the BB_WATCH bad battery seems most reasonable to me. May be it is that bad that it has just power for a few hours. And when I redid the test it was charged enough for the test.

Wim

[Wim Van den Wyngaert]

I left the station power off during my (sunny) skiing holiday (10 days).
It booted without date prompt.

I will test it again during the next time change (28 oct).

Wim