Synchronize newsgroups
actsync [-b hostid] [-d hostid] [-g max] [-i ignore_file]
[-I hostid] [-k] [-l hostid] [-m] [-n name]
[-o fmt] [-p min_%_unchg] [-q hostid] [-s size]
[-s spool_dir] [-t hostid] [-T] [-v verbose_lvl]
[-z sec] [host1] host2
actsyncd [-x] actsync.cfg [debug_level [debug_outfmt] ]
Actsync(8) permits one to synchronize, compare or merge two active(5) files. With this utility one may add, change or remove newsgroups on the local news server to make it similar to the list the newsgroups found on another system or file. The synchronization need not be exact. Local differences in newsgroup lists may be maintained and preserved. Certain newsgroup errors may be detected and optionally corrected.
There are several reasons to run actsync(8) (or actsyncd(8)), on a periodic basis. Among the reasons are:
A control message to add, change or remove a newsgroup may fail to reach your site.
Your control.ctl(5) is out of date or incomplete.
News articles for a new newsgroup arrive ahead (sometimes days ahead) of the control message.
Control messages may be forged, thus bypassing the restrictions found in control.ctl(5).
Your active(5) file may have been trashed.
If either host1 or host2 begin with a ``.'' or ``/'', then they assumed to be a name of a file containing information in the active(5) format. The getlist(1) utility may be used to obtain copy a remote system's active file. Newsgroup information from a file may be treated as if it was obtained from a host. In this man page host1 and host2 are called hosts, even though they may be file names.
If a host argument does not begin with ``.'' or ``/'', is assumed to be a hostname or Internet address. In this case, actsync(8) will attempt to use the NNTP protocol to obtain a copy of the the specified system's active file.
Regardless how the active file information is obtained, the actions of actsync(8) remain the same.
If only one host is specified, it is assumed to be host2. If host1, is not specified, it assumed to be the default local NNTP server as specified by the NNTPSERVER environment variable, or by the server value found in inn.conf(5).
The newsgroup synchronization by default, involves all newsgroups found on both hosts. One may also synchronize on a subset of newsgroups by directing actsync(8) to ignore certain newsgroups from both systems.
The actsyncd(8) daemon provides a convenient interface to configure and run actsync(8). If a host is not initially reachable, the daemon will thrice retry 3 times, waiting 5 minutes before each set of 3 retries. This daemon runs in the foreground, sending output to standard output and standard error.
If the -x flag is given to actsyncd(8), then a ctlinndxexec will be used instead of a ctlinndreload to load the newly modified active file.
The configuration filename for the daemon is given in the actsync.cfg argument. The actsync.cfg file is required to have 4 lines:
host=host2 spool=/var/spool/news ignore_file=ignore_file flags=actsyncd (8) options
The keyword must start at the beginning of the line, and there may be no whitespace before the ``='' character. Blank lines are ignored. Comments start with ``#'' and are ignored. All other lines may produce undefined results.
The host config file line refers to the host2 value to sync off of. The spool config file lines determines where top the news spool tree is to be found. The ignore_file config file line names the ignore file to be used by actsync(8). The flags config file line refers to all flags that you wish to pass to actsync(8).
Note that the -i ignore_file option, the -o format option and the -S spool_dir option should not be given in the flags= line because they are automatically taken care of by actsyncd(8).
The options to actsync(8) are as follows:
-b hostid
This flag causes actsync(8) to ignore newsgroups with ``bork.bork.bork'' style names. That is, newsgroups whose last 3 components are identical. For example, the following newsgroups have bork style names:
alt.helms.dork.dork.dork alt.auto.accident.sue.sue.sue alt.election.vote.vote.vote
The value hostid determines on which hosts this action is performed:
0 neither host 1 local default server 2 remove server 12 both servers 21 both servers
The default is -b 0, no bork newsgroups are ignored.
-d hostid
This flag causes actsync(8) to ignore newsgroups that have all numeric path components. The hostid value is interpreted the same as in -b. For example, the following newsgroups have numeric path components:
alt.prime.chongo.23209 391581.times.2.to_the.216193.power.-1 99.bottles.of.treacle.on.the.wall linfield.class.envio_bio.101.d
The newsgroups directory of a newsgroups with a all numeric component could conflict with an article from another group. For example, the directory for the first newsgroup listed above is the same path as article number 23209 from the newsgroup:
alt.prime.chongo
The default is -d 0, all numeric newsgroups from both hosts will be processed.
-g max
Ignore any newsgroup with more than max levels. For example, -g 6 would ignore:
alt.feinstien.votes.to.trash.freedom.of.speech alt.senator.exon.enemy.of.the.internet alt.crypto.export.laws.dumb.dumb.dumb
but would not ignore:
alt.feinstien.acts.like.a.republican alt.exon.admendment alt.crypto.export.laws
If max is 0, then the max level feature is disabled.
By default, the max level feature is disabled.
-i ignore_file
The ignore_file allows one to have a fine degree of control over which newsgroups are ignored. It contains a set of rules that specifies which newsgroups will be checked and which will be ignored.
By default, these rules apply to both hosts. This can be modified by using the -I hostid flag.
By default, all newsgroups are checked. If no ignore_file if specified, or if the ignore file contains no rule lines, all newsgroups will be checked.
Blank lines, and text after a ``#'' are considered comments and are ignored.
Rule lines consist of tokens separated by whitespace. Rule lines may be one of two forms:
c newsgroup [type ...] i newsgroup [type ...]
If the rule begins with a c then the rule requests certain newsgroups to be checked. If the rule begins with an i then the rule requests certain newsgroups to be ignored. The newsgroup field may be a specific newsgroup, or a wildmat(3) pattern.
If one or more types are specified, then the rule applies to the newsgroup only if is of the specified type. Types refer to the 4th field of the active(5) file. A type may be one of:
y n m j x =group.name
Unlike active files, the group.name may be a newsgroup name or a wildmat(3) pattern. Also, ``='' is equivalent to ``=*''.
For given rule line may, one may not repeat a given pattern type. For example, one may not have more than one type that begins with ``='', per line. However, one may achieve the effect of multiple ``='' types by using multiple rule lines for the same group.
By default, all newsgroups are candidates to be checked. If an ignore file is used, each newsgroup in turn is checked against the ignore file. If multiple lines match a given newsgroup, the last line in the ignore file is used.
For example, consider the following ignore file lines:
i *.general c *.general m i nsa.general
The newsgroup: ba.general would be ignored if it was not moderated. The newsgroup: mod.general would be checked if it was moderated. The newsgroup: nsa.general would be ignored even if it was moderated.
-I hostid
This flag restricts which hosts, the ignore file applies. The hostid value is interpreted the same as in -b.
This flag may be useful in conjustion with the -m merge flag. For example:
actsync -i actsync.ign -I 2 -m host1 host2
will keep all newsgroups currently on host1. It will also will only compare host1 groups with non-ignored newsgroups from host2.
The default is -I 12, newsgroups from both hosts to be ignored per the -I hostid flag.
-k
By default, any newsgroup on host1 that is in error will be considered for removal. This causes actsync(8) simply ignore such newsgroups. This flag, in combination with -m will prevent any newsgroup from being scheduled for removal.
-l hostid
Flag problem newsgroups of type ``='' from host1 or host2 as errors. The hostid value is interpreted the same as in -b. Newsgroups of type ``='' are newsgroups active entries that have 4th field that begins with ``=''. I.e., a newsgroup that is equivalent to another newsgroup.
A newsgroup that is equivalent to itself, or that is in a equivalence chain that loops around to itself is a problem. A newsgroup that is in a chain that is longer than 16 is a problem group. A newsgroup that is equivalent to a non-existent newsgroup is a problem. A newsgroup that is equivalent to a newsgroup that is has a error of some kind a problem. However, a newsgroup that is equivalent to an ignored newsgroup is not a problem.
By default, problem newsgroups from both hosts are marked as errors.
-m
Merge newsgroups instead of sync. By default, if a newsgroup exists on host1 but not host2, it will be scheduled to be removed. This flag disables this process, permitting newsgroups unique to host1 to be kept.
-n name
Newsgroups that are created, are created via the ctlinnd(8) command. By default, the creator name used is actsync. This flag changes the creator name to name.
-o fmt
Determine the output / action format of this utility. The fmt may one of:
a output in active(5) format, a1 output in active(5) format, and output host1 non-error ignored groups ak output in active(5) format, but use host2 hi & low (2nd & 3rd active fields) values for any newsgroup being created aK output in active(5) format, but use host2 hi & low (2nd & 3rd active fields) values for all newsgroups found in host2 a1k output in active(5) format, but use host2 hi & low (2nd & 3rd active fields) values for any newsgroup being created, and output host1 non-error ignored groups a1K output in active(5) format, but use host2 hi & low (2nd & 3rd active fields) values for all newsgroups found in host2, and output host1 non-error ignored groups ak1 same as a1k aK1 same as a1K c output in ctlinnd(8) format x no output, directly exec ctlinnd(8) commands xi no output, directly exec ctlinnd(8) commands, in an interactive mode
The a, a1, ak, aK, a1k, a1K, ak1 and aK1 style formats allow one to form a new active file instead of producing ctlinnd(8) commands. They use hi & low values of 0000000000 and 0000000001 respectively for newsgroups that are created. The ak and aK variants change the the hi & low (2nd & 3rd active fields). In the case of ak, newsgroups created take their hi & low values from host2. In the case of aK, all newsgroups found on host2 take their hi & low values from host2.
The c format produces ctlinnd(8) commands. No actions are taken because actsync(8) simply prints ctlinnd(8) commands on standard output. The sync (or merge if -m) with host2 may be accomplished by piping this output into sh(1). A paranoid person might prefer to use x or xi in case a newsgroup name or type contains bogus characters that might be interpreted by sh(1). Even so, this output format is useful to let you see how host1 may be synced (or merge) with host2.
The sync (or merge if -m) may be accomplished directly by use of the x. With this format, actsync(8) uses the execl(2) system call to directly executes ctlinnd(8) commands. Because of the exec, there is no risk of bogus newsgroups containing bogus characters causing a shell to do bogus (or dangerous) things. The output of such execs may be seen of the verbosity level is at least 2.
The actsync(8) utility will pause for 4 seconds before each command is executed if -o x is selected. See the -z sec flag below.
The xi format interactively prompts on standard output and reads directives on standard input. One may pick and choose changes using this format.
Care should be taken when producing active(5) formatted output. One should check to be sure that actsync(8) exited with a zero status prior to using such output. Also one should realize that such output will not contain lines ignored by the -i ignore_file process even if -p 100 is used.
By default, -o c is assumed.
-p min_%_unchg
By default, the actsync(8) utility has safeguards against performing massive changes. If fewer than min_%_unchg percent of the non-ignored lines from host1 remain unchanged, no actions (output, execution, etc.) are performed and actsync(8) exits with a non-zero exit status. The min_%_unchg may be a floating point value such as 66.666.
A change is considered a host1 line that was found to be in error, was removed, was added or was changed. Changing the 2nd or 3rd active fields via -oak or -o aK are not considered changes by -p.
To force actsync(8) to accept any amount of change, use the -p 0 option. To force actsync(8) to reject any changes, use the -p 100 option.
Care should be taken when producing active(5) formatted output. One should check to be sure that actsync(8) exited with a zero status prior to using such output. Also one should realize that such output will not contain lines ignored by the -i ignore_file process even if -p 100 is used.
By default, 96% of the lines not ignored in host1 must be unchanged. That is, by default, -p 90 is assumed.
-q hostid
By default, all newsgroup errors are reported on standard errors. This flag quiets errors from host1 or host2. The hostid value is interpreted the same as in -b.
-s size
If size>0, then ignore newsgroups with names longer than size, and ignore newsgroups equivalenced to names longer than size. Length checking is perform on both the local and remote hosts.
By default, size is 0 and thus no length checking is performed.
-S spool_dir
For each new newsgroup (i.e., selected groups found on host2 that were not found on host), the newsgroup directory under spool_dir will be examined. If this newsgroup directory exists, then the hi & low (2nd & 3rd active fields) values of the active entry will be changed to reflect the range articles found.
This flag is only useful with -o a, -o a1, -o ak, -o aK, -o alk, -o alK, -o ak1 or -o aK1. The -S spool_dir will override any hi & low (2nd & 3rd active fields) values that would normally have been used. This is an important and very much recommended option as it will prevent article number collisions on newsgroups that have been removed previous but still have unexpired articles in them.
-t hostid
Ignore improper newsgroups with only a top component from host1 or host2. The hostid value is interpreted the same as in -b. The following newsgroups are considered proper newsgroups for top only names:
control general junk test to
For example, the following newsgroup names are improper because they only contain a top level component:
dole_for_pres dos microsoft windoes95
By default, all improper top level only newsgroups from the remote ( -t 2 ) are ignored.
-T
This flag causes host2 newsgroups from new hierarchies to be ignored. Normally if only host2 has the newsgroup chongo.was.here then it will be created for host1. However if host1 does not have any 'chongo.*' newsgroups and this flag is given, then chongo.was.here will be ignored and will not be created on host1.
-v verbose_lvl
No default, actsync(8) is not verbose. This flag controls the verbosity level as follows:
0 no debug or status reports (default) 1 print summary, if work was needed or done 2 print actions, exec output & summary, if work was needed or done 3 print actions, exec output & summary 4 full debug output
-z sec
If -o x is selected, actsync(8) will pause for sec seconds before each command is executed. This helps prevent innd(8) from being busyed-out if a large number of ctlinnd(8) commands are needed. One can disable this sleeping by using -z 0.
By default, actsync(8) will pause for 4 seconds before each command is executed if -o x is selected.
Determine the difference (but don't change anything) between your newsgroup set and uunet's set:
actsync news.uu.net
Same as above, with full debug and progress reports:
actsync -v 4 news.uu.net
Force a site to have the same newsgroups some other site:
actsync -o x master
This may be useful to sync a slave site to its master, or to sync internal site to a gateway.
Compare your site with uunet, disregarding local groups and certain local differences with uunet. Produce a report if any differences were encountered:
actsync -v 2 -i actsync.ign news.uu.net
where actsync.ign contains:
# Don't compare to.* groups as they will differ. # i to.* # These are our local groups that nobody else # (should) carry. So ignore them for the sake # of the compare. # i nsa.* # These groups are local favorites, so keep them # even if uunet does not carry them. # i ca.dump.bob.dorman i ca.keep.bob.dorman i alt.tv.dinosaurs.barney.die.die.die i alt.tv.dinosaurs.barney.love.love.love i alt.sounds.* =alt.binaries.sounds.*
To interactively sync against news.uu.net, using the same ignore file:
actsync -o xi -v 2 -i actsync.ign news.uu.net
Based on newsgroups that you decided to keep, one could make changes to the actsync.ign file:
# Don't compare to.* groups as they will differ. # i to.* # These are our local groups that nobody else # (should) carry. So ignore them for the sake # of the compare. # i nsa.* # These groups are local favorites, so keep them # even if uunet does not carry them. # i ca.dump.bob.dorman i alt.tv.dinosaurs.barney.die.die.die i alt.sounds.* =alt.binaries.sounds.* # Don't sync test groups, except for ones that are # moderated or that are under the gnu hierarchy. i *.test c *.test m # check moderated test groups c gnu.*.test c gnu.test # just in case it ever exists
Automatic processing may be setup by using the following actsync.cfg file:
# host to sync off of (host2) host=news.uu.net # location of the ignore file ignore_file=/usr/local/news/actsync.ign # where nwes articles are kept spool=/var/spool/news # actsync(8) flags # # Automatic execs, report if something was done, # otherwise don't say anything, don't report # uunet active file problems, just ignore # the effect entries. flags=-o x -v 2 -q 2
and then by running:
actsyncd /usr/local/news/actsync.cfg
One may produce a trial actsyncd(8) run without changing anything on the server by supplying the debug_level arg:
actsyncd /usr/local/news/actsync.cfg 2
The debug_level causes actsyncd(8) to run actsync(8) with an -v debug_level (overriding any -v flag on the flags line), prevents any changes from being made to the active(5) file, writes a new active file to standard output and writes debug messages to standard error.
If the debug_outfmt arg is also given to actsyncd(8) then the data written to standard output will be in -o fBdebug_outfmt instead of in -o a1 format. The following /bin/sh command:
actsyncd /usr/local/news/actsync.cfg 4 c >cmd 2>dbg
Will operate in debug mode, not change the active(5) file, write ctlinnd(8) style commands to cmd and write debug statements to dbg.
To check only the major hierarchies against news.uu,net, use the following actsync.ign file:
# by default, ignore everything i * # check the major groups c comp.* c gnu.* c sci.* c alt.* c misc.* c news.* c rec.* c soc.* c talk.*
and running:
actsync -i actsync.ign news.uu.net
To determine the differences between your old active and your current default server:
actsync /usr/local/news/active.old -
To report but not fix any newsgroup problems with the current active file:
actsync - -
To detect any newsgroup errors on your local server, and to remove any *.bork.bork.bork style silly newsgroup names:
actsync -b 2 - -
The active file produced by:
actsync ... flags ... -o x erehwon.honey.edu
or by:
actsync ... flags ... -o c erehwon.honey.edu | sh
is effectively the same as the active file produced by:
ctlinnd pause 'running actsync' rm -f active.new actsync ... flags ... -o a1 erehwon.honey.edu > active.new rm -f active.old ln active active.old mv active.new active ctlinnd reload active 'running actsync' ctlinnd go 'running actsync'
It should be noted that the above 'pause', 'actsync', 'reload' and 'go' method is faster. However, in order to avoid article number collisions on newsgroups that have been removed previous but still have unexpired articles in them, it is very much recommended that the -S spool_dir be used with any of the -oa* flags. Thus, a much better and safer version of the above would be:
ctlinnd pause 'running actsync' rm -f active.new actsync ... flags ... -o a1 -S /var/spool/news erehwon.honey.edu > active.new rm -f active.old ln active active.old mv active.new active ctlinnd reload active 'running actsync' ctlinnd go 'running actsync'
The above process is similar to what actsyncd(8) does by default.
Careless use of this tool may result in the addition change or removal of newsgroups that you don't want. You should avoid using the x output format until you are sure it will do what you want.
A number of innd(8) servers will corrupt the active file when lots of newgroups and rmgroups are performed. This is not a actsync(8) bug, it is a server bug. Using the pause, actsync, reload and go method noted above is much more likely to avoid this problem.
If a newsgroup appears multiple times, actsync(8) will treat all copies as errors. However, if the group is marked for removal, only one rmgroup will be issued.
The timeout for ctlinnd(8) commands is fixed at 30 seconds when running in ``x'' or ``xi'' output format. Perhaps the timeout value should be controlled via a command line option?
Written by Landon Curt Noll <[email protected]> for InterNetNews.