1 INFO-VAX	Fri, 05 Nov 2004	Volume 2004 : Issue 615       Contents:% Re: DS10L SCSI card - recommendations ? Re: OpenVMS 7.2 , 'Secure' FTP Transmissions to Non-VMS Partner ? Re: OpenVMS 7.2 , 'Secure' FTP Transmissions to Non-VMS Partner   Re: Read DCL output to RDB table' Re: Request for feedback - COPY utility ' Re: Request for feedback - COPY utility ' Re: Request for feedback - COPY utility ( Re: Samba 2.2.8 for OpenVMS, new release( Re: Samba 2.2.8 for OpenVMS, new release Re: Volume shadowing stalls...  F ----------------------------------------------------------------------  $ Date: Fri, 5 Nov 2004 20:51:40 +1100# From: "Gremlin" <not-here@all.mate> . Subject: Re: DS10L SCSI card - recommendations/ Message-ID: <418b4d2e$1@duster.adelaide.on.net>    Thanks  F I will try it soon - I am using a Seagate 60Gb IDE drive and, despite I comments in the group, it is working fine as a web server hosting to the  I world.  No I/O problems or throughput problems and it receives about 17K  	 hits/day.   H Most interesting is watching robots and real (?) hackers as they try to / figure out what it is and what to do with it :)   9 "Rodman S. Regier" <rsr@hfx.andara.com> wrote in message  2 news:rc7jo0prp2a29a4ptdhr48eiv8tc9cgvkg@4ax.com...C > On Sat, 16 Oct 2004 13:16:19 +1000, "Gremlin" <not-here@all.mate>  > wrote: > L >>Rather than using IDE (perhaps slow as in an earlier thread), could I haveF >>recommendations for a SCSI card for a 36/72Gb disk - third party if 
 >>possibleK >>as I want to get one in the next day or so and can't access "real" Compaq  >>cards.  Adaptec? > @ > My sources indicate that the 3X-KZPEA-DB and the Adaptec 39160" > start life from the same source. > F > I have successfuly used 39160's in DS10L's and booted 3rd party SCSIG > U320 disks.  I'm using it with external disks, but you should be able + > to use with a single internal SCSI drive.  >  >  >  > B > ................................................................; >       Posted via TITANnews - Uncensored Newsgroups Access 3 >             >>>> at http://www.TitanNews.com <<<< B > -=Every Newsgroup - Anonymous, UNCENSORED, BROADBAND Downloads=- >    ------------------------------  % Date: Fri, 05 Nov 2004 22:25:36 +0800  From: prep@prep.synonet.com H Subject: Re: OpenVMS 7.2 , 'Secure' FTP Transmissions to Non-VMS Partner- Message-ID: <87654k74pr.fsf@prep.synonet.com>   $ "BillC" <bclark_at_lrgh.org> writes:  B > greetings - Healthcare Environment, regularly 'share' files withB > Government and Private (Commercial) healthcare organizations. AmE > anticipating a requirement to 'SECURE' data transmissions and would C > like to know if OpenVMS tools capable of Secure transmissions are 3 > available.  Thanks in advance for any information   C VMS Encryption. Remind hp that health care is supposed to be one of D the prefered markets and update it now that DES is depreciated would& be a good move. If they care, that is.   --  < Paul Repacholi                               1 Crescent Rd.,7 +61 (08) 9257-1001                           Kalamunda. @                                              West Australia 6076* comp.os.vms,- The Older, Grumpier Slashdot. Raw, Cooked or Well-done, it's all half baked.F EPIC, The Architecture of the future, always has been, always will be.   ------------------------------  % Date: Fri, 05 Nov 2004 19:09:13 +0100  From: Dirk Munk <munk@home.nl>H Subject: Re: OpenVMS 7.2 , 'Secure' FTP Transmissions to Non-VMS Partner2 Message-ID: <cmgfk9$57q$1@news4.zwoll1.ov.home.nl>  S There is SFTP in the SSH package of TCP/IP 5.4 . But then you need VMS 7.3.x AFAIK.    BillC wrote:M > greetings - Healthcare Environment, regularly 'share' files with Government 
 > and Private I > (Commercial) healthcare organizations. Am anticipating a requirement to 
 > 'SECURE'N > data transmissions and would like to know if OpenVMS tools capable of Secure > transmissions are available.' > Thanks in advance for any information  >  >    ------------------------------   Date: 5 Nov 2004 03:09:34 -0800 * From: javiercoego@gmail.com (Javier Coego)) Subject: Re: Read DCL output to RDB table = Message-ID: <d78fdf9b.0411050309.322212d2@posting.google.com>     - Thanks a lot, Bart and Phil. -  & - Both ways, RDO and RMU works well. -  B - I coded a solution in C language, invoking SQL sentences via SQLE Module Procedures, but I have to admit that your solutions are better  than mine. -  E - Anyway I attach sqlmod and C files, and the way to create the final  exe file. -   < - Bart, eights years later, your procedure still works. - :)   - Thanks, Javi. -    =============================   # miModulo.sqlmod  (spec module file)    MODULE MI_MODULO
 DIALECT SQL92 
 LANGUAGE C AUTHORIZATION SCIDES21 ALIAS RDB$HANDLE PARAMETER COLONS  D DECLARE ALIAS FOR FILENAME D_USUARIOS:[PAS.SCIDES21.BBDD]ADMIN_ADMIN   PROCEDURE SET_TRANS 
 (SQLCODE);$ SET TRANSACTION READ WRITE RESERVING  ORD_USU_VMS FOR EXCLUSIVE WRITE;   PROCEDURE INSERT_DATA  (SQLCODE :USU_RECORD RECORD P_TIPO_ORDEN CHAR(1),  P_PARAMS CHAR(512) END RECORD);   INSERT INTO ORD_USU_VMS  (TIPO_ORDEN,PARAMS)  VALUES1 (:USU_RECORD.P_TIPO_ORDEN, :USU_RECORD.P_PARAMS);    PROCEDURE COMMIT_TRANS
 (SQLCODE); COMMIT;    ==========================   miSql.c   (C file)   #include <stdio.h> #include <stdlib.h>  #include <string.h>  #include <sql_rdb_headers.h>   struct t_misDatos {     char orden[2];     char parametros[513]; };   struct t_misDatos misDatos;    int return_status;  
 long sqlcode;    void SET_TRANS (long *sqlcode); B void INSERT_DATA (long *sqlcode, struct t_misDatos *misDatos_ptr);" void COMMIT_TRANS (long *sqlcode);  ! int main(int argc, char *argv[]){       misDatos.orden[0] = '\0';!    misDatos.parametros[0] = '\0';   #    strcat(misDatos.orden, argv[1]); (    strcat(misDatos.parametros, argv[2]);      misDatos.orden[1] = '\0';#    misDatos.parametros[512] = '\0';       SET_TRANS(&sqlcode);     if (sqlcode != 0) {1       printf ("\Start of transaction failed.\n");        sql_signal();     }%    INSERT_DATA (&sqlcode, &misDatos);     if (sqlcode != 0) {$       printf ("\nInsert failed.\n");       sql_signal();     }    COMMIT_TRANS(&sqlcode);    if (sqlcode != 0) {0       printf ("\nEnd of transaction failed.\n");       sql_signal();     }    return 0; }    ==============================   CREATE EXE FILE   , sqlmod miModulo.sqlmod    (==> miModulo.obj) cc miSql.c		(==>miSql.obj)> DEFINE LNK$LIBRARY SYS$LIBRARY:SQL$USER71.OLB     (for RDB7.1,) SQL$USER.OLB if previous versions or RDB) - link miSql.obj, miModulo.obj		(==> miSql.exe) 2 mySymbol:==$DGA3:[USUARIOS.PAS.SCIDES21.BBDD]miSql mySymboluno dos        lederman@star.enet.dec.DISABLE-JUNK-EMAIL.com (Bart Z. Lederman) wrote in message news:<Y0rid.2303$o51.1534@news.cpqcorp.net>...G > Years ago, as a customer, I used to use command procedures to monitor C > network performance and load the numbers into an Rdb database for B > later analysis (and to spot problems).  I did it with DCL, whichG > created a command procdure that did an RDO STORE.  If I did it again, E > I'd probably use SQL syntax instead, but the principle is the same. > > This gets around some of the problems you seem to be seeing. > D > As an example, I'm attaching one such procedure.  (Note, I haven'tD > tried this procedure for more than 8 years, so there may have beenD > some syntax changes in the Rdb product since then.)  It's probablyB > more complicated (parses out more values) than you need for your/ > application, but others might find it useful.  > / > $ WRITE SYS$OUTPUT F$ENVIRONMENT("PROCEDURE") ? > $! Parse out counts on Ethernet device and store in database.  > $! > $ ON ERROR THEN CONTINUE > $! > $ quote = """  > $! > $ IF F$MODE () .EQS. "BATCH" > $ THEN) > $    SET PROCESS /NAME = "QNA-0 Counts"  > $ ELSE' > $    ON CONTROL_Y THEN $ GOTO exitout 	 > $ ENDIF  > $!1 > $! Find out if we have the required privileges.  > $!' > $ savepriv = F$SETPRV ("OPER,SYSPRV") 4 > $ IF F$PRIVILEGE ("OPER,SYSPRV") THEN GOTO gotpriv > $!H > $ WRITE SYS$OUTPUT "You do not have the correct privilege to do this." > $ EXIT > $! > $ gotpriv: > $!( > $! save repeat time interval, if given > $! > $ interval = P1  > $! > $! Define the NCP command  > $! > $ NCP :== $NCP > $! > $ bigloop: > $!1 > $ NCP SHOW LINE QNA-0 COUNT TO qna_0_counts.tmp  > $!> > $ OPEN /WRITE /ERROR = nofile outfile qna_0_counts_store.com > $!( > $ WRITE outfile "$ RUN SYS$SYSTEM:RDO"D > $ WRITE outfile "invoke database filename 'user0:[accounting]qna'"V > $ WRITE outfile "start_transaction read_write reserving qna_counts for shared write"/ > $ WRITE outfile "store D in qna_counts using"  > $! > $! Get fields  > $!7 > $ OPEN /READ /ERROR = nofile tmpfile qna_0_counts.tmp  > $! > $ readloop1: > $!< > $ ON ERROR THEN $ GOTO endfile1			! reset after goto error > $!* > $ READ /END = endfile1 tmpfile tmpstring! > $ length = F$LENGTH (tmpstring) B > $ IF length .LT. 15 THEN $ GOTO readloop1	! some lines too short > $!: > $ count = F$EXTRACT (0, 12, tmpstring)			! count portionE > $ IF count .EQS. "            " THEN $ GOTO readloop1	! blank field & > $ count = F$EDIT (count, "COLLAPSE")D > $ IF count .EQS. "LineCounter" THEN $ GOTO date_time	! date string > $ length = length - 14> > $ field = F$EXTRACT (14, length, tmpstring)		! type of count4 > $ field = F$EDIT (field, "COLLAPSE, TRIM, UPCASE") > $!L > $ IF F$LOCATE (">", count) .NE. F$LENGTH (count) THEN -	! counter overflowA >         count = F$EXTRACT (1, F$LENGTH (count), count)	! remove  > $!B > $ IF F$LOCATE (",", field) .NE. F$LENGTH (field)	! remove commas > $ THEN > $    tmpstring = fieldH > $    field = F$ELEMENT (0, ",", tmpstring) + F$ELEMENT (1, ",", field)	 > $ ENDIF  > $!A > $ IF F$LOCATE (":", field) .NE. F$LENGTH (field)	! remove colon  > $ THEN( > $    field = F$ELEMENT (0, ":", field)	 > $ ENDIF  > $!8 > $ ON WARNING THEN $ GOTO goto_error	! in case of error > $!* > $ GOTO 'field'				! process this counter > $! > $ goto_error:  > $ ON ERROR THEN $ CONTINUE- > $ WRITE SYS$OUTPUT "Unknown field " + field  > $ GOTO readloop1:  > $! > $ date_time: > $ length = length - 20- > $ count = F$EXTRACT (20, length, tmpstring) D > $ WRITE outfile "    d.date_time = " + quote + count + quote + ";" > $ GOTO readloop1 > $! > $ SECONDSSINCELASTZEROED: D > $ WRITE outfile "    d.SECONDS_SINCE_LAST_ZEROED = " + count + ";" > $ GOTO readloop1 > $! > $ DATABLOCKSRECEIVED: ? > $ WRITE outfile "    d.DATA_BLOCKS_RECEIVED = " + count + ";"  > $ GOTO readloop1 > $! > $ MULTICASTBLOCKSRECEIVED:D > $ WRITE outfile "    d.MULTICAST_BLOCKS_RECEIVED = " + count + ";" > $ GOTO readloop1 > $! > $ RECEIVEFAILURE:  > $ RECEIVEFAILUREINCLUDING:: > $ WRITE outfile "    d.RECEIVE_FAILURE = " + count + ";" > $ GOTO readloop1 > $! > $ BYTESRECEIVED:8 >  WRITE outfile "    d.BYTES_RECEIVED = " + count + ";" > $ GOTO readloop1 > $! > $ MULTICASTBYTESRECEIVED: C > $ WRITE outfile "    d.MULTICAST_BYTES_RECEIVED = " + count + ";"  > $ GOTO readloop1 > $! > $ DATAOVERRUN:7 > $ WRITE outfile "    d.DATA_OVERRUN = " + count + ";"  > $ GOTO readloop1 > $! > $ DATABLOCKSSENT: ; > $ WRITE outfile "    d.DATA_BLOCKS_SENT = " + count + ";"  > $ GOTO readloop1 > $! > $ MULTICASTBLOCKSSENT:@ > $ WRITE outfile "    d.MULTICAST_BLOCKS_SENT = " + count + ";" > $ GOTO readloop1 > $!! > $ BLOCKSSENTMULTIPLECOLLISIONS: J > $ WRITE outfile "    d.BLOCKS_SENT_MULTIPLE_COLLISIONS = " + count + ";" > $ GOTO readloop1 > $! > $ BLOCKSSENTSINGLECOLLISION:G > $ WRITE outfile "    d.BLOCKS_SENT_SINGLE_COLLISION = " + count + ";"  > $ GOTO readloop1 > $!  > $ BLOCKSSENTINITIALLYDEFERRED:I > $ WRITE outfile "    d.BLOCKS_SENT_INITIALLY_DEFERRED = " + count + ";"  > $ GOTO readloop1 > $! > $ BYTESSENT:5 > $ WRITE outfile "    d.BYTES_SENT = " + count + ";"  > $ GOTO readloop1 > $! > $ MULTICASTBYTESSENT: ? > $ WRITE outfile "    d.MULTICAST_BYTES_SENT = " + count + ";"  > $ GOTO readloop1 > $! > $ SENDFAILURE: > $ SENDFAILUREINCLUDING: 7 > $ WRITE outfile "    d.SEND_FAILURE = " + count + ";"  > $ GOTO readloop1 > $!  > $ COLLISIONDETECTCHECKFAILURE:I > $ WRITE outfile "    d.COLLISION_DETECT_CHECK_FAILURE = " + count + ";"  > $ GOTO readloop1 > $!! > $ UNRECOGNIZEDFRAMEDESTINATION: I > $ WRITE outfile "    d.UNRECOGNIZED_FRAME_DESTINATION = " + count + ";"  > $ GOTO readloop1 > $! > $ SYSTEMBUFFERUNAVAILABLE:D > $ WRITE outfile "    d.SYSTEM_BUFFER_UNAVAILABLE = " + count + ";" > $ GOTO readloop1 > $! > $ USERBUFFERUNAVAILABLE:< > $ WRITE outfile "    d.USER_BUFFER_UNAVAILABLE = " + count > $!( > $! When we hit this field, we're done. > $!
 > $ endfile1:  > $! > $ WRITE outfile "end_store"  > $ WRITE outfile "commit" > $ WRITE outfile "exit" > $ WRITE outfile "$ EXIT" > $ CLOSE outfile  > $ CLOSE tmpfile $ > $ DELETE /NOLOG qna_0_counts.tmp;* > $! > $ NCP ZERO LINE QNA-0 COUNT  > $!) > $! Now store the counts in the database  > $! > $ @qna_0_counts_store.com  > $! > $! Don't need file anymore.  > $!* > $ DELETE /NOLOG qna_0_counts_store.com;* > $!1 > $! Check to see if we were asked to auto-repeat  > $!* > $ IF interval .EQS. "" THEN GOTO exitout > $ WAIT 'interval'  > $ GOTO bigloop > $! > $ exitout: > $!" > $ savepriv = F$SETPRV (savepriv) > $! > $ nofile:  > $! > $ EXIT $STATUS > $! > $!: > $! B. Z. Lederman	17-Oct-1991	Store date in Rdb database >  >  >  > --  * >  B. Z. Lederman   Personal Opinions Only > : >  Posting to a News group does NOT give anyone permission: >  to send me advertising by E-mail or put me on a mailing >  list of any kind. > 7 >  Please remove the "DISABLE-JUNK-EMAIL" if you have a 7 >  legitimate reason to E-mail a response to this post.    ------------------------------  # Date: Fri, 05 Nov 2004 09:09:35 GMT # From: "Phil" <dooleys@snowy.net.au> 0 Subject: Re: Request for feedback - COPY utility= Message-ID: <jrHid.15156$K7.15036@news-server.bigpond.net.au>   A "Andy Bustamante" <a_c_bustamante@earthlink.net> wrote in message 8 news:aZzid.38647$QJ3.36829@newssvr21.news.prodigy.com... > H > We use copy every night.  On the majority of my sites, I create a diskL > backup ($ copy my$database:*.dat /excl=big_static_file.dat my$backup:*.* )I > In  large cases this takes 30-45 minutes and adds to a downtime window.  The H > backup files are then used to create a tape backup.  This reduces down timeJ > and reduces the number of disks that BACKUP does i/o on.  In my example,0 > my$database is defined to span multiple disks.L We use backup to create savesets on disk and then copy to tape the next day.B The backup to disk used to be a lot quicker than copy, though this
 difference+ may be less with more recent disk hardware.  Phil   ------------------------------  % Date: Fri, 05 Nov 2004 04:42:32 -0500 - From: JF Mezei <jfmezei.spamnot@teksavvy.com> 0 Subject: Re: Request for feedback - COPY utility, Message-ID: <418B4ADD.DCCA3778@teksavvy.com>  K While trying to figure out some logging issues for a DSL modem behind a NAT : router that does the pesky PPPoE thing, I came to think...  T How about COPY/UPDATE which would use logic similar to a volume shadowing operation:  L compare blocks of target file and only actually write modified/added blocks.  H For speed, I assume some form of windowing where you asynchronously readN multiple blocks and compare them once a pair has been fetched (and then either' trigger the ansync write if necessary).   , ok, so now I must return to reading manuals.   ------------------------------   Date: 5 Nov 2004 09:31:10 -0800 , From: JimStrehlow@data911.com (Jim Strehlow)0 Subject: Re: Request for feedback - COPY utility< Message-ID: <4b6ec350.0411050931.eebd1be@posting.google.com>  b "Guy Peleg" <guy.peleg@remove_this_hp.com> wrote in message news:<418a0010@usenet01.boi.hp.com>...3 ...performance improvements for the copy utility...    Can you increase7  - the largest block size previously created by $BACKUP G    Currently I can only $COPY a backup saveset up to 32256 blocks large   1  - the /NETWORK_BLOCK_COUNT to be larger than 127   ? Let us know what we need to change for the OpenVMS user quotas.   . Thank you for your efforts to improve OpenVMS.  8 Jim Strehlow, Data911 OpenVMS and Database Administrator Alameda, CA, USA   ------------------------------  * Date: Fri, 5 Nov 2004 13:09:58 +0000 (UTC) From: david20@alpha1.mdx.ac.uk1 Subject: Re: Samba 2.2.8 for OpenVMS, new release ) Message-ID: <cmfu36$9gi$1@news.mdx.ac.uk>   ` In article <418AF899.55CDC557@comcast.net>, David J Dachtera <djesys.nospam@comcast.net> writes: >"John E. Malmberg" wrote:	 >> [snip]  > F >I sense this is on the border of devolving into a "religious war", so% >let me sum it up as simply as I can:  > H >At some point , the Samba team will have to "put a stake in the ground"A >and say, "HPSS (HP Secure Samba, or whatvere it might be called) I >requires OpenVMS Vx.y-z or later", just as the SWS team did. That pretty , >well solves the great bulk of the problems. > A >I still maintain, and will not be swayed from, the position that B >requiring installation by building from source, and requiring theH >presence of a prescribed build environment on the customers' machine to> >install Samba from source is wholly, entirely and irrevocablyH >unacceptable, and making such a demand will render the entire effort an >abysmal failure.  > G >Linking from object at install time is a proven, reliable distribution B >method. Contact Process Software and solicit their experience forF >specifics. (BEFORE you click that <Reply> button, remember: "requires >OpenVMS Vx.y-z or later".)  >   O Process seems to have done reasonably well in supporting the latest versions of ' their products on older versions of VMS     2 PMDF 6.2       -   VAX or ALPHA VMS 6.1 or greater  , TCPWARE 5.6    -   VAX VMS 5.5-2 and later  *                    Alpha VMS 6.1 and later  , MULTINET 5.0   -   VAX VMS 5.5-2 and later  *                    Alpha VMS 6.1 and later    N Compare this with DEC TCPIP services which seems to be tied relatively closely to the version of VMS.  0 Dec TCPIP services 5.4   -   VMS 7.3-1 and 7.3-2      
 David Webb Security team leader CCSS Middlesex University   ------------------------------  % Date: Fri, 05 Nov 2004 12:05:40 -0500 - From: "John E. Malmberg" <wb8tyw@qsl.network> 1 Subject: Re: Samba 2.2.8 for OpenVMS, new release 1 Message-ID: <YoOdncdtbPt5LxbcRVn-qw@adelphia.com>    David J Dachtera wrote:  > "John E. Malmberg" wrote:  >  >>[snip] > G > I sense this is on the border of devolving into a "religious war", so & > let me sum it up as simply as I can: > I > At some point , the Samba team will have to "put a stake in the ground" B > and say, "HPSS (HP Secure Samba, or whatvere it might be called)J > requires OpenVMS Vx.y-z or later", just as the SWS team did. That pretty- > well solves the great bulk of the problems.   H Currently the only "Samba team" is the one at http://www.samba.org, and B they do not currently include OpenVMS in their active development.  G When and if HP offers a Samba based product for OpenVMS, then they may  E determine what versions of SAMBA on what versions of OpenVMS will be  
 supported.  A HP-UX supports their own distribution of SAMBA, and maintains it  G independent of the "Samba team".  I do not know more details than that   about the HP-UX distribution.   B > I still maintain, and will not be swayed from, the position thatC > requiring installation by building from source, and requiring the I > presence of a prescribed build environment on the customers' machine to ? > install Samba from source is wholly, entirely and irrevocably I > unacceptable, and making such a demand will render the entire effort an  > abysmal failure.    G No one ever suggested that any HP supported distribution would only be  E supplying source.  I would expect that it would be the same as other  * open source products that they distribute.  E In my case for my hobby builds, and that is my personal choice, I am  B only distributing source kits.  And I will state my reasons again.  A 1. Those distributions have only been test built on OpenVMS Alpha >     8.2 EFT at the time that I post them on Encompasserve.org.  A 2. It is for mostly untested.  I do not have the facilities to do $     anything more than simple tests.  > 3. 2.2.12 only has minimal improvements over the 2.2.8 releaseA     that binaries are available for.  That is, unless you want to @     test SAMBA on OpenVMS I64.  In which case, I can not produce!     binaries for that now anyway.   A 4. It is intended for use only by those sites that are able to do E     their own builds and tests and to help diagnose and fix problems.   H > Linking from object at install time is a proven, reliable distributionC > method. Contact Process Software and solicit their experience for G > specifics. (BEFORE you click that <Reply> button, remember: "requires  > OpenVMS Vx.y-z or later".)  C The only reason to require a link at install time is if there is a  > component that is linked against the system image.  The 2.2.x H executables have this requirement.  Otherwise, only binaries need to be  distributed.  G By getting SAMBA on OpenVMS to use the "Persona System Services", that  E restriction will be removed, and pure executables will be able to be  8 distributed, which should result in smaller binary kits.  I However binaries targeted for OpenVMS 8.3 may have significantly  better  E performance than binaries targeted for OpenVMS 7.3 as a hypothetical  H example.  The binaries targeted at OpenVMS 7.3 are likely to still work 8 on OpenVMS 8.3, but may not give as good of performance.   -John  wb8tyw@qsl.network Personal Opinion Only    ------------------------------  # Date: Fri, 05 Nov 2004 18:14:57 GMT 1 From: Keith Parris <keithparris_NOSPAM@yahoo.com> ' Subject: Re: Volume shadowing stalls... 2 Message-ID: <BqPid.2442$AR2.1826@news.cpqcorp.net>   Hal Kuff wrote: N > Sometimes we get very very long merge operations that seem to take days evenK > though the system is quiet.... I think there is a back-off sensitivity to 3 > systenm I/O that is perhaps getting triggered....   1 In the original design for Volume Shadowing (see  I http://www.research.compaq.com/wrl/DECarchives/DTJ/DTJ301/DTJ301SC.TXT),  H it was felt that the impact of merge I/Os by the Shadow_Server would be G the primary detrimental effect on application I/O performance during a  G Merge. As a result, Shadowing was designed to throttle back merge I/Os  D in the presence of application I/Os. The thresholds also seem to be E implemented in such a way that with today's fast I/O configurations,  ? this defensive reaction is often triggered too readily, so the  H recommendation seems to be that the SHAD$MERGE_DELAY* logical names may  be needed to get good behavior.   I In my studies recently, it appears the worst impact during Merges is not  E the merge I/Os, but the need for repeated merges of areas with "hot"  I read activity that are ahead of the merge fence. And throttling back the  D merge I/Os just prolongs the agony, as it slows the progress of the ) merge fence across the disk surface. See  6 http://www2.openvms.org/kparris/s2003_volshad_perf.ppt  F Of course, the best solution would be for you to upgrade to 7.3-2 and ; install the Host-Based Mini-Merge (HBMM) kit for Shadowing.    ------------------------------   End of INFO-VAX 2004.615 ************************