SGI Networker 4.2.9

home *** CD-ROM | disk | FTP | other *** search

/ SGI Networker 4.2.9 / Networker 4.2.9 for IRIX 6.2, 6.3 and 6.4.img / relnotes / networker / ch5.z / ch5

Wrap

Text File | 1997-07-08 | 33KB | 727 lines

- 1 - 5. _K_n_o_w_n__P_r_o_b_l_e_m_s__a_n_d__W_o_r_k_a_r_o_u_n_d_s The problems in this chapter are listed in relative order of importance, starting with the most important. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Be sure to look at SYSLOG and daemon.log for important NetWorker messages. Many important NetWorker messages will be written to the /usr/var/SYSLOG or to /nsr/logs/daemon.log files. It is critically important that the system administrator watch these logs closely to learn about errors or other problems encountered by NetWorker. If you use the System Log Viewer, then choose Warning from the Filter menu to display NetWorker root messages. _I_n_c_i_d_e_n_t #_2_2_1_4_2_9 SSSSuuuummmmmmmmaaaarrrryyyy:::: If your licenses expire, configuration is partially destroyed. _I_m_p_o_r_t_a_n_t! The IRIX NetWorker product is very sensitive to invalid licenses licenses! Although backed-up data will not be lost if licenses become invalid, configuration information can be lost. Also, your site can be prey to this bug even if your licenses are not invalid-if your system clock is ever set to an incorrect value, your system may believe that your licenses are invalid even if they are not. To protect against the possibility of lost configuration information, be sure to save a copy of the files stored in the /_n_s_r/_r_e_s directory. For instance, you could put the following in a script called /_u_s_r/_e_t_c/_n_s_r._p_r_o_t_e_c_t: ####!!!! ////bbbbiiiinnnn////sssshhhh iiiiffff [[[[ !!!! ----dddd ////nnnnssssrrrr////rrrreeeessss....bbbbaaaacccckkkkuuuupppp ]]]];;;; tttthhhheeeennnn \\\\ mmmmkkkkddddiiiirrrr ----pppp ////nnnnssssrrrr////rrrreeeessss....bbbbaaaacccckkkkuuuupppp;;;; ffffiiii ccccdddd ////nnnnssssrrrr....rrrreeeessss ccccpppp $$$$iiii ////nnnnssssrrrr////rrrreeeessss....bbbbaaaacccckkkkuuuupppp Then whenever you want to save your IRIX NetWorker resource files (for instance, right after you make important changes to them), you would type: #### ////uuuussssrrrr////eeeettttcccc////nnnnssssrrrr....pppprrrrooootttteeeecccctttt - 2 - _I_n_c_i_d_e_n_t #_2_2_2_8_0_5 SSSSuuuummmmmmmmaaaarrrryyyy:::: IRIX NetWorker will allow by default any server to access files of any client. _S_e_c_u_r_i_t_y _W_a_r_n_i_n_g! IRIX NetWorker as installed by _i_n_s_t on NetWorker client systems will allow any NetWorker server (Silicon Graphics or other) access to all of the files on the client machine. If you have any internal security concerns, insert an additional argument into the invocation of the _n_s_r_e_x_e_c_d command in the file /_e_t_c/_i_n_i_t._d/_n_e_t_w_o_r_k_e_r. This line usually reads: ////uuuussssrrrr////eeeettttcccc////nnnnssssrrrreeeexxxxeeeeccccdddd >>>>////ddddeeeevvvv////ccccoooonnnnssssoooolllleeee 2222>>>>&&&&1111 For more secure installations, it should read: ////uuuussssrrrr////eeeettttcccc////nnnnssssrrrreeeexxxxeeeeccccdddd ----ssss nnnnssssrrrr----sssseeeerrrrvvvveeeerrrr----nnnnaaaammmmeeee >>>>////ddddeeeevvvv////ccccoooonnnnssssoooolllleeee 2222>>>>&&&&1111 You can have as many ----ssss [[[[nnnnaaaammmmeeee]]]] arguments as you want-and nnnnssssrrrr----sssseeeerrrrvvvveeeerrrr----nnnnaaaammmmeeee should be the name of your NetWorker server. In our case, the line reads: ////uuuussssrrrr////eeeettttcccc////nnnnssssrrrreeeexxxxeeeeccccdddd ----ssss bbbbiiiittttbbbbuuuucccckkkkeeeetttt....eeeennnnggggrrrr....ssssggggiiii....ccccoooommmm \\\\ >>>>////ddddeeeevvvv////ccccoooonnnnssssoooolllleeee 2222>>>>&&&&1111 For more information, see the _n_s_r_e_x_e_c_d(1M) reference page. _I_n_c_i_d_e_n_t #_2_5_4_1_5_9 SSSSuuuummmmmmmmaaaarrrryyyy:::: IRIX NetWorker will allow the user account _o_p_e_r_a_t_o_r access to files of any NetWorker client. _S_e_c_u_r_i_t_y _W_a_r_n_i_n_g! A user account named _o_p_e_r_a_t_o_r is able to read and/or write any file on a NetWorker client system, independent of its UID, using the Backup and Recover windows. NetWorker checks for that account name and bypasses some security checks for that particular account. This feature supports sites that have true operators that mount tapes and recover files for users. These operators are able to write to any file on the system. This can be used to gain _r_o_o_t access. WWWWoooorrrrkkkkaaaarrrroooouuuunnnndddd:::: Ensure that no NetWorker client system (including the NetWorker server system) has an account named _o_p_e_r_a_t_o_r. _I_n_c_i_d_e_n_t #_2_2_3_3_9_2 SSSSuuuummmmmmmmaaaarrrryyyy:::: IRIX NetWorker abandons saveset, then is unable to save later on. - 3 - If your IRIX NetWorker client machines' filesystems are too large, or the network is not working properly, the timeout value in some of the NNNNSSSSRRRR GGGGrrrroooouuuupppp resources may be too small. As installed, this timeout is set to 30 minutes. Silicon Graphics has found that by increasing it to 60 minutes, all but the most intractable examples of this problem are solved. To increase the time, follow the instructions in the _I_R_I_X _N_e_t_W_o_r_k_e_r _A_d_m_i_n_i_s_t_r_a_t_o_r'_s _G_u_i_d_e on using the nnnnwwwwaaaaddddmmmmiiiinnnn tool. In nnnnwwwwaaaaddddmmmmiiiinnnn, you will need to select _C_u_s_t_o_m_i_z_e->_G_r_o_u_p_s from the menu bar, which will create a window. In that window's menu bar, select _V_i_e_w->_D_e_t_a_i_l_s. Now select the name of the problematic group in the upper part of the window. Scroll down in the lower part of the window until you see the field iiiinnnnaaaaccccttttiiiivvvviiiittttyyyy ttttiiiimmmmeeeeoooouuuutttt. Set its value to 60. Then click the _A_p_p_l_y button to commit the change. _I_n_c_i_d_e_n_t #_2_3_2_7_4_2 SSSSuuuummmmmmmmaaaarrrryyyy:::: NetWorker as non-_r_o_o_t fails... IRIX NetWorker backups and restores may fail when run by a user other than _r_o_o_t. If you see a blank status window when you try to use _n_w_b_a_c_k_u_p, and you are running as a user other than _r_o_o_t, try running as _r_o_o_t. The same goes for other IRIX NetWorker commands, but this is the most common occurrence of this bug. _I_n_c_i_d_e_n_t #_2_5_2_8_0_5 SSSSuuuummmmmmmmaaaarrrryyyy:::: Adding devices is not working correctly. Adding devices to IRIX NetWorker does not work if there are no devices currently configured. Also, attempting to change the definition of an existing device does not work if there is only one device defined. When adding or changing devices, ensure that you do not delete the last one or attempt to change it. Instead, create and configure a new device, then delete the old one. If your IRIX NetWorker site is licensed for only one device, be sure to delete the old device before rebooting the system. _I_n_c_i_d_e_n_t #_2_1_4_3_4_9 SSSSuuuummmmmmmmaaaarrrryyyy:::: Saves don't follow symbolic links. - 4 - IRIX 5.xxxx and IRIX 6.xxxx use many symbolic links, and IRIX NetWorker will not follow them. For example, an IRIX NetWorker site might set up a client system so that the /_u_s_r/_m_a_i_l partition is backed up. Under IRIX 4.xxxx, /_u_s_r/_m_a_i_l was a directory in the /_u_s_r partition, but under IRIX 5.xxxx, /_u_s_r/_m_a_i_l is a symbolic link to /_v_a_r/_m_a_i_l. This means that in this example, the files in the /_v_a_r/_m_a_i_l directory would not be backed up despite their apparent explicit inclusion in the requested savesets for the client system. WWWWoooorrrrkkkkaaaarrrroooouuuunnnnddddssss:::: +o Configure so that as many client machines as possible save the special AAAAllllllll saveset. +o Ensure that all explicitly saved filesystems are really filesystems, and not symbolic links. _I_n_c_i_d_e_n_t #_2_2_3_3_8 SSSSuuuummmmmmmmaaaarrrryyyy:::: NetWorker wants to use _D_e_f_a_u_l_t _P_o_o_l when not present or referenced. _W_a_r_n_i_n_g: In certain cases, IRIX NetWorker does not select the correct PPPPoooooooollll to which to send a client index save. In these cases, the index file backups go to the _D_e_f_a_u_l_t _P_o_o_l. This is a problem when sites are not using the _D_e_f_a_u_l_t _P_o_o_l at all. Index save requests can prevent the next backup of the same backup GGGGrrrroooouuuupppp. WWWWoooorrrrkkkkaaaarrrroooouuuunnnnddddssss:::: +o IIIIffff yyyyoooouuuu hhhhaaaavvvveeee aaaa ttttaaaappppeeee rrrroooobbbbooootttt, ensure that at all times at least one appendable tape that is labeled _D_e_f_a_u_l_t exists in the tape robot. +o IIIIffff yyyyoooouuuu hhhhaaaavvvveeee aaaa ssssiiiinnnngggglllleeee ttttaaaappppeeee ddddrrrriiiivvvveeee, (not a tape robot), use the _D_e_f_a_u_l_t _P_o_o_l for your backups to avoid this problem. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: NetWorker client software for old systems doesn't look right. The graphical interfaces supplied for use on IRIX 4.xxxx do not correctly use the SGI X11 defaults information to set color schemes and fonts. WWWWoooorrrrkkkkaaaarrrroooouuuunnnndddd:::: Upgrade to IRIX 5.3 or later. - 5 - _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: The Exabyte 10i jukebox has a bug which affects IRIX NetWorker operations. The internal firmware in an EXB-10i jukebox has a bug that causes it to not respond correctly after the front door to the jukebox has been opened and closed. The next jukebox operation after the door has been closed will receive a SCSI-bus timeout after 30 seconds and fail. The next jukebox operation after that should work correctly. The suggested workaround is to perform a manual jukebox operation (for example, _n_s_r_j_b -_H) immediately afterward if the jukebox door has to be opened and closed. The manual operation will receive an error, but the next automatic jukebox operation will perform correctly. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Using Hewlett Packard C-1533A 4mm tape drives with your IRIX system. Users who wish to use HP-C1533 4mm tape drives, or robotic units which include them (such as the Spectra Logic SL-40000 jukebox) must insert these lines in their /_v_a_r/_s_y_s_g_e_n/_m_a_s_t_e_r._d/_s_c_s_i file after the entry for the Archive Python drive: {{{{ DDDDAAAATTTTTTTTAAAAPPPPEEEE,,,, TTTTPPPPDDDDAAAATTTT,,,, 2222,,,, 6666,,,, """"HHHHPPPP"""",,,, """"CCCC1111555555553333AAAA"""",,,, 0000,,,, 0000,,,, {{{{0000,,,, 0000,,,, 0000,,,, 0000}}}},,,, MMMMTTTTCCCCAAAANNNN____BBBBSSSSFFFF||||MMMMTTTTCCCCAAAANNNN____BBBBSSSSRRRR||||MMMMTTTTCCCCAAAANNNN____AAAAPPPPPPPPEEEENNNNDDDD||||MMMMTTTTCCCCAAAANNNN____SSSSEEEETTTTMMMMKKKK|||| MMMMTTTTCCCCAAAANNNN____PPPPAAAARRRRTTTT||||MMMMTTTTCCCCAAAANNNN____PPPPRRRREEEEVVVV||||MMMMTTTTCCCCAAAANNNN____SSSSYYYYNNNNCCCC||||MMMMTTTTCCCCAAAANNNN____SSSSPPPPEEEEOOOODDDD|||| MMMMTTTTCCCCAAAANNNN____CCCCHHHHKKKKRRRRDDDDYYYY||||MMMMTTTTCCCCAAAANNNN____VVVVAAAARRRR||||MMMMTTTTCCCCAAAANNNN____SSSSEEEETTTTSSSSZZZZ||||MMMMTTTTCCCCAAAANNNN____SSSSIIIILLLLIIII|||| MMMMTTTTCCCCAAAANNNN____SSSSEEEEEEEEKKKK||||MMMMTTTTCCCCAAAANNNN____CCCCHHHHTTTTYYYYPPPPEEEEAAAANNNNYYYY,,,, 44440000,,,, 8888****66660000,,,, 4444****66660000,,,, 5555****66660000,,,, 555511112222,,,, 555511112222****111122228888 }}}},,,, After that, rebuild the kernel and reboot: #### aaaauuuuttttooooccccoooonnnnffffiiiigggg ----ffff #### rrrreeeebbbbooooooootttt The switch settings on HP-C1533 drives as set from the factory will not work with your Silicon Graphics system. You may find the bank of eight switches on the bottom of the drive. The set of switch settings SGI recommends for these drives is switches 1, 2, and 5 on, all others off. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Using Hewlett-Packard C-1553A 4mm tape robotic units with your IRIX system. - 6 - Users who wish to use HP-C1553 robotic units must insert these lines in their /_v_a_r/_s_y_s_g_e_n/_m_a_s_t_e_r._d/_s_c_s_i file after the entry for the Archive Python drive: {{{{ DDDDAAAATTTTTTTTAAAAPPPPEEEE,,,, TTTTPPPPDDDDAAAATTTT,,,, 2222,,,, 6666,,,, """"HHHHPPPP"""",,,, """"CCCC1111555555553333AAAA"""",,,, 0000,,,, 0000,,,, {{{{0000,,,, 0000,,,, 0000,,,, 0000}}}},,,, MMMMTTTTCCCCAAAANNNN____BBBBSSSSFFFF||||MMMMTTTTCCCCAAAANNNN____BBBBSSSSRRRR||||MMMMTTTTCCCCAAAANNNN____AAAAPPPPPPPPEEEENNNNDDDD||||MMMMTTTTCCCCAAAANNNN____SSSSEEEETTTTMMMMKKKK|||| MMMMTTTTCCCCAAAANNNN____PPPPAAAARRRRTTTT||||MMMMTTTTCCCCAAAANNNN____PPPPRRRREEEEVVVV||||MMMMTTTTCCCCAAAANNNN____SSSSYYYYNNNNCCCC||||MMMMTTTTCCCCAAAANNNN____SSSSPPPPEEEEOOOODDDD|||| MMMMTTTTCCCCAAAANNNN____CCCCHHHHKKKKRRRRDDDDYYYY||||MMMMTTTTCCCCAAAANNNN____VVVVAAAARRRR||||MMMMTTTTCCCCAAAANNNN____SSSSEEEETTTTSSSSZZZZ||||MMMMTTTTCCCCAAAANNNN____SSSSIIIILLLLIIII|||| MMMMTTTTCCCCAAAANNNN____SSSSEEEEEEEEKKKK||||MMMMTTTTCCCCAAAANNNN____CCCCHHHHTTTTYYYYPPPPEEEEAAAANNNNYYYY,,,, 44440000,,,, 8888****66660000,,,, 4444****66660000,,,, 5555****66660000,,,, 555511112222,,,, 555511112222****111122228888 }}}},,,, After that, rebuild the kernel and reboot: #### aaaauuuuttttooooccccoooonnnnffffiiiigggg ----ffff #### rrrreeeebbbbooooooootttt The switch settings on HP-C1553 robotic units as set from the factory will not work with your Silicon Graphics system. You may find the bank of eight option switches on the bottom of the unit. The switch settings SGI recommends for these drives is switches 1, 2, and 5 on, all others off. There are no user-settable switches on the unit which control robotic options. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Do not use the top slot in Exabyte 210 jukebox. This slot is ignored by NetWorker, so do not put a data tape in this slot. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Summary: Do not embed special characters in schedule names. IRIX NetWorker enforces restrictions on characters in schedule names to alphanumeric characters and '-'. If an existing NetWorker server that uses other than these characters in schedule names is upgraded to IRIX NetWorker, shedule names will be changed to replace these characters, and client configurations that used these schedules will be changed to use the Default schedule. The simplest solution is to rename the schedule before upgrading. To do this, stop NetWorker using "/etc/init.d/networker stop". Replace the schedule name every where it appears in the file /nsr/res/nsr.res with a new name that does not use special characters. Then upgrade to IRIX NetWorker. Otherwise, the schedule must be re- - 7 - created under IRIX NetWorker and the server client configurations changed to use the new schedule. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Deleting and re-adding jukebox destroys volume label and slot information. If you delete a jukebox under nwadmin, it removes all the state and configuration associated with this jukebox, including volume labels and slot assignments. Do not delete a jukebox unless removing it permanently from the NetWorker server. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Server per-client index is limited to 4 GB (without PerformancePak) or 16 GB (with PerformancePak). The size of the IRIX NetWorker 4.2.x client index in the server directory ////nnnnssssrrrr////iiiinnnnddddeeeexxxx is limited to either 4 GB or 16 GB. Once this limit is reached, NetWorker notifies the system administrator and ceases to update the online index for this client during backups. Monitor index sizes in ////nnnnssssrrrr////iiiinnnnddddeeeexxxx and NetWorker notifications, and set browse policies to prevent per-client indexes from reaching 4 GB (or 16 GB if PerformancePak is installed). _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Licenses must be consistent; inconsistent licenses can cause incorrect behavior of the NetWorker server. Please be sure that your licenses are consistent. For example, there are three different base NetWorker licenses: the demo version, the Network Edition, and the Server Edition. (And there is also the NetWorker Edition upgrade from Server Edition.) Obviously, you only need one of these. But less obviously, you must not have more than one of these. It is a common assumption that the NetWorker server would require the Server Edition license, but this is usually incorrect. The Server Edition permits backups of only the server itself; most installations will use the Network Edition which allows multiple clients connect to the NetWorker server across the LAN. - 8 - If you have more than one jukebox, of different capacities, this is how to apply the licenses. First, add the licenses for all parts of NetWorker other than the jukeboxes to your license.dat file. Then, run nnnnssssrrrrccccaaaapppp ----vvvv. Then, add only the smallest jukebox license to your lllliiiicccceeeennnnsssseeee....ddddaaaatttt file, and run nnnnssssrrrrccccaaaapppp again. This will bind the small jukebox license to your smallest jukebox. Finally, add the larger jukebox license to your lllliiiicccceeeennnnsssseeee....ddddaaaatttt file, and again run nnnnssssrrrrccccaaaapppp. Repeat these last two steps for each additional jukebox license, one at a time, in order from smallest to largest. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Recovery of the NetWorker server after a crash can be time-consuming. If NetWorker's operations should be suddenly interrupted, as by a crash or other malfunction of the NetWorker server, it is possible that NetWorker's indexes may become corrupted. In this case, NetWorker must perform a consistency check of the indexes, and possibly recover them using any of a variety of techniques, depending on the severity of the damage. The time required depends on the speed of your CPU/disk subsystem, overall system loading, the overall size of the index, the actual data stored in the index, as well as possibly other factors. Following are some suggestions for minimizing the time that the NetWorker server requires to restart after a crash. The NetWorker server stores its file indexes in ////nnnnssssrrrr////iiiinnnnddddeeeexxxx////{{{{nnnnaaaammmmeeee----ooooffff----cccclllliiiieeeennnntttt}}}}////ddddbbbb, and when this file becomes corrupted, NetWorker must rebuild parts of it. NetWorker will do this automatically when it restarts, or the system administrator can do this manually using nnnnssssrrrrcccckkkk ----FFFF. Some general rules that apply here are useful to keep in mind. +o You must always maintain adequate free space under ////nnnnssssrrrr. The ////nnnnssssrrrr hierarchy must nnnneeeevvvveeeerrrr be allowed to run out of space; otherwise you risk index corruption. From a practical standpoint, this means putting ////nnnnssssrrrr in its own filesystem. +o If the NetWorker server crashes (usually because the entire machine crashes), then upon restarting the system, you must allow NetWorker to completely process - 9 - the indexes and verify their validity. Do not under any circumstances terminate any NetWorker processes while index processing is underway. (You can recognize this state by noting the CPU time used by nnnnssssrrrriiiinnnnddddeeeexxxxdddd.) +o While NetWorker is busy in this state of index recovery, then all other NetWorker operations will be locked out. This includes both saves and recoveries. DDDDoooo nnnnooootttt attempt to terminate any NetWorker processes in this situation; you'll just be right back where you started if you do. You must wait for the index recovery operation to complete. +o Do whatever you can to minimize the risk of abrupt NetWorker shutdown (crash). For instance, you should do whatever is necessary to ensure that ////nnnnssssrrrr never runs out of space. Using appropriate RAID technology to protect the ////nnnnssssrrrr hierarchy from corruption caused by disk failure is also recommended. +o Monitor the sizes of your indexes, because larger indexes will take significantly longer to validate after a NetWorker crash. You should establish an appropriate compromise between the recovery time and index size. As the NetWorker server restarts, each client index is processed in turn. For the fastest possible recovery time of the NetWorker server, the system administrator could process the indexes in parallel, using the following guidelines. (Note that this procedure is entirely optional; if you don't do this, then NetWorker will perform the equivalent operation, iterating serially over all the client indexes, as described above.) +o You must not allow the NetWorker daemons to start when the system comes up after the crash. +o Process each client index with nnnnssssrrrrcccckkkk ----FFFF {{{{cccclllliiiieeeennnntttt----nnnnaaaammmmeeee}}}}. You can run as many nnnnssssrrrrcccckkkk ----FFFF {{{{cccclllliiiieeeennnntttt----nnnnaaaammmmeeee}}}} processes as you like, each one operating on a different client index; they will all run in parallel. +o The free disk space requirement under ////nnnnssssrrrr will be highest using this approach. The rule of thumb is that you must have free space under ////nnnnssssrrrr equal to the combined sizes of all client indexes that are being processed simultaneously by nnnnssssrrrrcccckkkk, plus a 30% cushion. Of course, this could add up to a substantial requirement for free disk space, but it is the solution - 10 - that offers the fastest possible recovery time. +o Once all client indexes have been processed by nnnnssssrrrrcccckkkk, then the NetWorker daemons can be restarted for normal operations. +o It is probable that not all client indexes will need recovery. The nnnnssssrrrrcccckkkk process will notice which ones need recovery; for indexes which are not corrupted, nnnnssssrrrrcccckkkk will simply terminate. +o BBBBeeeeffffoooorrrreeee running nnnnssssrrrrcccckkkk manually, the system administrator mmmmuuuusssstttt ensure that if nnnnssssrrrrcccckkkk is interrupted (for instance, by a system crash), that the system won't automatically restart NetWorker's daemons until after the index recovery process is fully complete. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Index extent too small error message. If you should see an error message of this form: nnnnssssrrrrcccckkkk:::: WWWWAAAARRRRNNNNIIIINNNNGGGG:::: iiiinnnnddddeeeexxxx ////iiiinnnnddddeeeexxxx////nnnnssssrrrr////iiiinnnnddddeeeexxxx////<<<<cccclllliiiieeeennnntttt>>>>////ddddbbbb eeeexxxxtttteeeennnntttt #### ttttoooooooo ssssmmmmaaaallllllll ((((8888111199991111)))) it means that this original index conversion process was not performed correctly. The solution is to shut down NetWorker, and then use nnnnssssrrrrcccckkkk ----FFFF on the affected client. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Issues with machine alias names in NetWorker 4.2.5 or 4.2.9. NetWorker 4.2.x is much fussier than 4.1.3 with respect to the aliases you use for your machine names. In 4.2.x you must use exact (fully qualified) alias names. If you don't do this, you'll typically see messages of the form "usr root on machine fred is not on fred.sgi.com's access list". This message is generated by authentication logic that was introduced in 4.2.x, which is completely different than the logic used by 4.1.x. With NetWorker 4.2.x, a machine name (as it known to Networker) MUST be on the alias list. For example, if Networker knows a machine by the name "fred" and in the alias list there is only one alias "fred.sgi.com", the save will fail and it will issue something like "user root on machine fred is not on fred.sgi.com's access list". For this to work, the aliases must match, and "fred" is not a match for "fred.sgi.com". - 11 - When you create client entries using NetWorker 4.2.x, it automatically creates both long and short (i.e., FQDN and canonical) hostnames. So this issue should never be a concern unless you have changed things around, or have upgraded from a previous version of IRIX NetWorker. Of course, another approach is to ensure that the correct clients mentioned in the appropriate remotes access list. _N_o _I_n_c_i_d_e_n_t # SSSSuuuummmmmmmmaaaarrrryyyy:::: Clients can't have similar names. (One is a substring of the other.) NetWorker has trouble with clients that have similar names. Specifically, if you have one client whose name is a substring of the other client, this issue can arise. For example, your clients might be named fred and fred2. We're working to provide a solution for this at a future date.