Date: Fri, 1 Mar 2002 07:56:14 -0800
From: Russell E Owen
Subject: Re: 2.5m obslog 20020228 (MJD 52334)
>2.  PR3135 test output
>
>First I did:
>track/pterr=(noobj,nocor) 120,55 obs /rotty=obj /rota=0
>
>Then I did:
>track 120,60 mount
>0 0 I TCCStatus="TTT","NNN"; TCCPos=119.53,55.75,93.49;
>AxePos=119.53,55.75,93.4
>9
>track 120,60 mount
>0 0 I AxisErrCode="OK","OK","OK"
>track 120,60 mount
>0 0 I SpiderInstAng = 93.097572, 0.003091, 4521682944.53000
>track 120,60 mount
>0 4 > Started; Cmd='track 120,60 mount'
>0 4 W Text="one or more axes will be left halted"; AxisNoSlew=F,F,T;
>AxisErrCode
>="OK","OK","CannotCompute"
This is exactly right. The slew will happen, it's just warning that
the rotator cannot move. BBF (before bug fix) the slew would have
been entirely rejected.
>0 0 F Modu="prt_ReadReplySet"; Text="too many replies on port LTA19:"
>0 0 I Modu="prt_ReadReplySet"; Text="the following replies were read:"
>0 0 I Modu="prt_ReadReplySet"; Text="ERR: I don't have the semCmdPort
>semaphore"
Yes...well...it'd be nice to do this test with the semaphore.
Now that the TCC can report bad status, I propose we get rid of the
"does not have semaphore" message and rely on bad status to indicate
that the TCC can't move the telescope. It'd sure make testing a lot
easier.
>3. Pointing model fits
>
>...
>Sky RMS =   1.59
>Popn SD =   1.76
Congratulations. That's very nice!
>4. informTCCOfGProbes errors
>
>  From TCC:
>
>0 0 F Modu="blk_LoadInst"; Text="data for guide probe 1 incomplete or
>defective; file rejected"
>0 0 I Modu="blk_LoadInst"; Text="file:"
Looks like the word "file:" was sent to the TCC. Eric: if you want,
send me the output you are trying to send to the TCC and I'll vet it
for you. (You also have an example of good data from an earlier email
message; I can find or reconstruct if it it'll be helpful).
>5.  Results of string of TCC "SHOW LOCK" commands
>
>When we issued a string of "SHOW LOCK" commands to the TCC, we saw this error
>message in the stream, which we are not sure was the expected error mode.
>
>0 12 : Cmd='show lock'
>0 0 W UserNum=3; MsgsLost=177; Text="output resumes"
>0 0 W UserNum=3; MsgsLost=1; Text="user mailbox full; no more errors
>reported for this user until output resumes"
That is perfectly reasonable. You flooded SOP or some other reader.
Serious errors are if:
- A process times out while writing to the message hub. This is an
outstanding PR that may most easily be solved by buying a faster TCC
(though having fewer users or optimizing TCC code would also probably
solve it). I was hoping that the slightly faster output of the new
messaging system might improve things, but it's not all that likely.
The message hub's speed is determined primarily by how many users
there are and efficiently it is coded. A secondary effect, of course,
is how much CPU time can be spared for it, and that should have
improved with the new messaging code. (I'd hoped to do even better by
using unformatted output, but found I could not get reasonable
behavior for displaying partial commands from interactive logins).
- Output terminates for some user -- and I mean the error message
must say "terminates" in it. This indicates a bug in the new message
handling code. Fortunately it was not seen.
I will mark both of these PRs as closed later today (if nobody has
beaten me to it; I'm working from home this morning and can't get to
the SDSS bug database from here). Thanks for testing these!
-- Russell