CVS log for loncom/Attic/lonc

Up to [LON-CAPA] / loncom

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

- removing the old lonc

- make connections using hostname

- better error messages when dyng
- perl 5.8 seems create noexistant clients, if the client isnt there just go about our buisness

- 5 minutes is too long
- 5 attempts isn't enough

- convert the rest of the initialization steps to use londtransaction so that it will error out on timeouts

Documentation updates

Information from "SessionOne.html" of the Guts manual.

I'm not sure all of this is going to live in lonc; some parts may move
elsewhere (load balancing probably belongs in the login routines), and
it all may be picked up and dropping into lonnet.pm if that makes more
sense. The main point is that it's now in POD format and the above can
be done with copy & paste.

In HUPSMAN - execdir was being inquired after first use... move it around
so that execdir is defined prior to use.

- added new command to lond
  sethost:machinenamewanted
     returns either ok or unable_to_set
  version:MY VERSION
     returns version:MY VERSION

- init method comatible with older 0.6ish inti sequence

- lonc properly uses sethost and version

- Machines with multiple domains now supporteda somewhat
http://bugs.loncapa.org/show_bug.cgi?id=1320

  - Requirements:
      hosts.tab must have several enetries one for each hostid/domain
          but the domain name and IP addresses must be the same
      The last entry of the host must correspond with the entries in loncapa.conf

  - Known isssues,
      1) CSTR space has no domain info in it, username conflicts can occur
      2) while you can add the Author role to user, if the role isn't the main domain they can't author


  - Tested areas
      1) Course creation
      2) DOCS mods on course in main/non-main domain
      3) problems in course in main/non-main domain

- timeouts are bad, kill the child off and hope for something better on restart

- turning on KEEPALIVE
- adding more info to some status messages

best wishes to all.

Bug #885
lonnet did not generate unique buffer filenames when updating both the
resource and the meta-file, following removal of sleep commands. Added
PID and $dumpcount to buffer file name.

Also added "sort" to buffer flushing when machine comes online again, so
that definitely time-order of commands gets preserved.

Show status in "ps"

loncapa_apache.conf should not be directly specified (and certainly not
before loncapa.conf)

Replaced access.conf with loncapa_apache.conf.

using LONCAPA::Configuration::read_conf
BUG 129

Fixed USR1 handler to:
1. Reset retry counters to zero rather than clearing counter hash.
2. On all counters which were >= maxchildretries, restart the lonc
   server for the associated host.

Conditionalize the debug log entry on entry to  HandleInput.

Use single fully blocking select for input, output and connection requests.

Add DoSelect sub to encapsulate select in the presence of signals.

Inserted support functions to allow restructuring the lonc main loop.
Note that no actual logic changes have been implemented with this
edit.

Put the small timeout back into the read select for now.

Add londtransaction sub and centralize all interactions with lond through it.
londtransaction is signal safe given the current set of signals which maybe
received meaningfully by lonc.

Corrected and simplified child management logic:
1. Removed SIGCHLD handler, made parent main loop
   just be a wait, rather than sleep.
2. Corrected book-keeping error in lonc's make_child_process
   which caused single child death to trigger attempts to
   restart all children...whether actually dead or not.

updating documentation

Less trashy, still kaputt (but not more than before, I hope)

Attempt to be able to close connections from inside of lonnet UNTESTED

Skip empty lines in broken hosts.tabs

Removing old garbage

Testing - skip servers that do not ping.

- added logging messages

- didn't mean to commit that one

- logging info now in

adding POD; removing void context map statements -Scott Harrison

GPL

No wonder subreply never returned anything useful

Init can also time out

Times out communication, or so one hopes.

lonc also does reporting

Attempting to officially hang up

Asymmetric traffic: always more to write than read

Hopefully fixed truncated data - no reason to stop sending if there was
no error.

Correct error message

adding in code to make sure that lonc and lond do not run as a user
other than 'www' -Scott

fixing the "already running" HUP error by unlinking the relevant
.pid file in /home/httpd/perl/logs -Scott

turns out there is no need for redefining the die function
after all -Scott

Make sure that it is not running into an endless loop if dying while dying
(for example, if logthis fails)

(somewhat unexpectedly) need to move subroutines to top of code
due to limits of perl program parser (otherwise received predeclaration
error) -Scott

added in exception handling to catch abnormal exiting through 'DIE'
and 'QUIT' signals -Scott

Check if already running to avoid two instances

Bug Fix for Maximum Retry Attempts

Put in additional sleep so that children do not die before reaper in place
(lonc)
Marked up logs
Put in report for after the lonc/lond start into loncron

Fixed weird globbing to send delayed messages
Changed format of delayed message file to have no server: - no need

Senddelayed moved from lonnet and loncron to lonc - not tested

Implementation of encryption
Different login mechanisms
IP Spoof Protection

LON-CAPA Communication Module

Initial revision