Page 1 of 1

Tesro's real-time clock not keeping time!!!???

Posted: Thu May 10, 2018 11:27 pm
by Irinikus
This Tezro of mine is sure keeping me busy with all of it's shite!

It's definitely not keeping time when it's unplugged, and I have the sneaky suspicion that it's not properly keeping time when it's powered down either.

I'm going to leave it in a powered down mode, with L1 running during the course of today to confirm this.

It must be the case of the machine having a flat a Dallas chip (hard to believe, as these machines are rather new in SGI terms), but could it be anything else?

Re: Tesro's real-time clock not keeping time!!!???

Posted: Fri May 11, 2018 12:49 am
by Trippynet
Rather new? The Tezros were first made back in 2002. A battery that could be up to 16 years old could quite easily have run out of charge by now.

Anyway, if it's losing time, the battery is almost certainly the cause.

Re: Tesro's real-time clock not keeping time!!!???

Posted: Fri May 11, 2018 12:54 am
by Irinikus
I suppose then it will be Dallas chip modding time for me then!

Under normal circumstances I wouldn't worry about date and time on such a machine, but this may (or absolutely will) effect the licenses on the machine and may be part of the problem concerning getting flame licensed on the machine.

According to uunix, the system host ID is stored in two places on the machine: The io9 card and the system board, when I type in lmhostid in the console, I get 1000518 as my host ID, how would I be able to tell if there are not maybe two lmhostid's present on the machine, which are in conflict, and how could I possibly fix this if there are?

Re: Tesro's real-time clock not keeping time!!!???

Posted: Fri May 11, 2018 2:34 am
by uunix
At the l1 prompt (ctl t) what is the date?

Re: Tesro's real-time clock not keeping time!!!???

Posted: Fri May 11, 2018 2:47 am
by jan-jaap
IIRC there are two batteries: a yellow 'snaphat' for the PROM settings and system time, and a Dallas battery backed NVRAM for the L1. They keep time independently. The most essential things like system serial number are replicated in both, so as long as you replace either before the other runs out you should be fine.

That said, the serial number you mention (1000518) sounds like the number found on the back side of the system. The lmhostid is a different number and I believe it starts with a 'd'. It does for the O350 series, at least. For most of the older systems it's the MAC address of the onboard LAN, which starts with 080069 for SGI.

The snaphat is a current product. I believe they stopped making the Dallas a year or two ago, you should still be able to find a new one. Be sure not to buy a 'second hand' one as the battery will be dead, and look for the manufacturing date stamp on any chips offered. If it's from 2015 or 2016 it should last another 10 years or so, by then a CR2032 hack may be needed.

Re: Tesro's real-time clock not keeping time!!!???

Posted: Fri May 11, 2018 4:13 am
by Irinikus
jan-jaap wrote:IIRC there are two batteries: a yellow 'snaphat' for the PROM settings and system time, and a Dallas battery backed NVRAM for the L1. They keep time independently. The most essential things like system serial number are replicated in both, so as long as you replace either before the other runs out you should be fine.

That said, the serial number you mention (1000518) sounds like the number found on the back side of the system. The lmhostid is a different number and I believe it starts with a 'd'. It does for the O350 series, at least. For most of the older systems it's the MAC address of the onboard LAN, which starts with 080069 for SGI.

The snaphat is a current product. I believe they stopped making the Dallas a year or two ago, you should still be able to find a new one. Be sure not to buy a 'second hand' one as the battery will be dead, and look for the manufacturing date stamp on any chips offered. If it's from 2015 or 2016 it should last another 10 years or so, by then a CR2032 hack may be needed.


Thanks Jan, which one supplies IRIX with it's date and time, as that's the one that's more than likely flat and would need to be replaced first?

EDIT:

I've found a place in china which can sell me both parts, new for a total of $13.00

I wonder if Flame is not able to see the discrepancy between these two clocks, hence the licensing issue?

The two culprits:
IMG_3591.JPG

IMG_3592.JPG


The lmhostID for the machine is definitely 1000518 according to IRIX:
IMG_3555.JPG

Re: Tesro's real-time clock not keeping time!!!???

Posted: Sat May 12, 2018 3:15 am
by uunix
At the l1 prompt
serial all

This should display something like..

Code: Select all


001c01-L1>serial all

Data                            Location      Value
------------------------------  ------------  --------
Local System Serial Number      NVRAM         P1000534
Reference System Serial Number  NVRAM         P1000534
Local Brick Serial Number       EEPROM        NAM668
Reference Brick Serial Number   NVRAM         NAM668


EEPROM      Product Name    Serial         Part Number           Rev  T/W
----------  --------------  -------------  --------------------  ---  ------
INTERFACE   WS_INT_53       NAM668         030_1881_006          A    00
IO9         IO9             NAM407         030_1771_005          A    00
ODYSSEY     ASTODYV10       MVR649         030_1826_001          B    00
SNOWBALL    no hardware detected
NODE        IP53_2CPU       NBC188         030_1869_001          C    00
IO DGHTR    CHWS_IO_DAUG    NAG645         030_1875_003          A    00

EEPROM     JEDEC-SPD Info           Part Number        Rev Speed  SGI
---------- ------------------------ ------------------ ---- ------ --------
DIMM 0     CE0000000000000027FF2500 M3 47L6510BT0-CA0   0B   10.0  N/A
DIMM 2     7F7FFE00000000001200313C CM2202B1            2     8.0  N/A
DIMM 4     no hardware detected
DIMM 6     no hardware detected
DIMM 1     CE0000000000000028411D01 M3 47L6510BT0-CA0   0B   10.0  N/A
DIMM 3     7F7FFE000000000012003140 CM2202B1            2     8.0  N/A
DIMM 5     no hardware detected
DIMM 7     no hardware detected


Notice the top lines
Local System Serial Number NVRAM P1000534
Reference System Serial Number NVRAM P1000534

Remove the 'P' and we have our hostid. Now, here we can see that both locations match. Also, this machine looks to be an early machine with a low 1000xxx number. This number does match the serial number sticker on the rear of the machine.

Re: Tesro's real-time clock not keeping time!!!???

Posted: Wed May 16, 2018 7:04 am
by Irinikus
I haven't taken a picture, but my hostid matches in both locations.

Which one of those locations refers to the io9 card with the snaphat?

Does the reference nvram keep a reference time for the system (to which the other time keeper is adjusted?), if this is the case, and the reference nvram happens to be on the io9 card, is it then really necessary to replace the Dallas chip, as snaphats are readily available?

I am able to source the required snapshot battery locally:
https://za.rs-online.com/web/p/battery- ... s/1891239/

I am able to source the Dallas chip on ebay from a company in china, but I'm sceptical about doing this, just in case they sell me a fake component, and as a result it causes damage to my system board: (the Dallas chip battery mod may be the best option here, as it will be a longer lasting solution)
https://www.ebay.co.uk/itm/1pcs-DS1742W ... 1438.l2649

If I were to carry out a Dallas chip battery mod, I would rather acquire a second chip do carry it out on, rather than messing with the one that's currently inside the machine.

I have the facility at work to X-ray the chip if required. (baggage X-ray machine)

Am I able to swap out the snaphat while the system is in a powered down mode, with L1 running in order to preserve the contents of the nvram on the io9 card?

I know that this doesn't sound like a good thing to do, but the nvram which it powers while the system is unplugged is on the io9 card, whereas the nvram on the system board is within the Dallas chip itself.

EDIT:

Removing the snaphat with L1 running probably wouldn't be a good idea, as it contains both a battery and the crystal needed to clock the surface-mounted SNAPHAT SOIC.
http://www.st.com/content/ccc/resource/ ... 149846.pdf