[OpenAFS] How to correct errors in prdb

D Brashear shadow@gmail.com
Sun, 23 Feb 2014 22:29:57 -0500


--001a1133352ef53aa604f31e95eb
Content-Type: text/plain; charset=ISO-8859-1

You can dump it with pt_util, and then reload it; editing of the raw file
can of course be done while it's in text format.


On Thu, Feb 20, 2014 at 4:53 AM, Kostas Liakakis <kostas@physics.auth.gr>wrote:

> Hello,
>
> Recently an aging AFS cell we had online for several years, started
> experiencing problems: ghost volumes not willing to be removed from vldb,
> volumes appearing with the same id but could be accessed by different
> names... we traced it down to a corrupted vldb file, we fixed it using
> vldb_check and it appears to behave again.
>
> While dealing with it, I also did a prdb_check and it came back with this:
>
> [root@srv-10-206-123-246 logs]# prdb_check -database prdb.DB0
> Bad user/group dicotomy in membership list
>   Entry at 722624: flags 0x2, id -269i, next 722816.
>   c:04/16/2006 22:02:02 a:06/28/2013 10:45:56 r:05/16/2013 14:38:43
> n:time-not-set
>   ids  10207  10718  10772  11072  11120  12059  12134  13108  13226  13247
>   hash (id 0 name 136256).  Owner 1i, creator 1i
>   quota groups 0, foreign users 0.  Mem: 57, inst: 0
>   Owned chain 1155392, next owned 715520, inst ptrs(0 0 0).
>   Name is 'webmasters'
>     Entry at 722816: flags 0x4, id -269i, next 1089728.
>     c:time-not-set   a:time-not-set   r:time-not-set   n:time-not-set
>     ids  13598  10611  11608  11137  12117  10027  11584  14027  12365
> -310
>          11123  12031  14013  11127  13010  14024  13241  12050  13728
>  13202
>          10927  14768  10001  13381  12478  14283  11944  12970  13323
>  15396
>          13197  10086  14770  11444  15628  11676  13216  14263  12686
> Entry membership count is inconsistent: 56 entries refer to this one
>   Entry at 722624: flags 0x2, id -269i, next 722816.
>   c:04/16/2006 22:02:02 a:06/28/2013 10:45:56 r:05/16/2013 14:38:43
> n:time-not-set
>   ids  10207  10718  10772  11072  11120  12059  12134  13108  13226  13247
>   hash (id 0 name 136256).  Owner 1i, creator 1i
>   quota groups 0, foreign users 0.  Mem: 57, inst: 0
>   Owned chain 1155392, next owned 715520, inst ptrs(0 0 0).
>   Name is 'webmasters'
>     Entry at 722816: flags 0x4, id -269i, next 1089728.
>     c:time-not-set   a:time-not-set   r:time-not-set   n:time-not-set
>     ids  13598  10611  11608  11137  12117  10027  11584  14027  12365
> -310
>          11123  12031  14013  11127  13010  14024  13241  12050  13728
>  13202
>          10927  14768  10001  13381  12478  14283  11944  12970  13323
>  15396
>          13197  10086  14770  11444  15628  11676  13216  14263  12686
>     Entry at 1089728: flags 0x4, id -269i, next 0.
>     c:time-not-set   a:time-not-set   r:time-not-set   n:time-not-set
>     ids  12841  14278  13208  15078  11212  15651  13259  12842
> Entry membership count is inconsistent: 2 entries refer to this one
>   Entry at 844544: flags 0x2, id -310i, next 0.
>   c:12/19/2007 15:14:51 a:12/19/2007 15:52:37 r:time-not-set n:time-not-set
>   ids  11123
>   hash (id 0 name 418688).  Owner -269i, creator 1i
>   quota groups 0, foreign users 0.  Mem: 1, inst: 1
>   Owned chain 0, next owned 795200, inst ptrs(0 -269 0).
>   Name is 'webmasters:alumni.physics.auth.gr'
>
> The cell has 4 prdb/vldb servers, 3 of them are version 1.4.5 and the last
> one, 1.6.6.
>
> The newer versioned server was installed just recently because the
> vldb_check from 1.4.5 could not fix the vldb errors. Along the same lines,
> prdb_check from 1.4.5 distribution does not report any errors on prdb,
> while the one from 1.6.6 gives the above mentioned errors.
>
> Though not alarming to my eyes at least and no evident problem on the cell
> so far, we would like to fix the prdb problems as well. Question is how?
> prdb_check utility is missing a -fix flag.
>
>
> Thanks for any input,
>
> -Kostas.
>
>
>


-- 
D

--001a1133352ef53aa604f31e95eb
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">You can dump it with pt_util, and then reload it; editing =
of the raw file can of course be done while it&#39;s in text format.<br></d=
iv><div class=3D"gmail_extra"><br><br><div class=3D"gmail_quote">On Thu, Fe=
b 20, 2014 at 4:53 AM, Kostas Liakakis <span dir=3D"ltr">&lt;<a href=3D"mai=
lto:kostas@physics.auth.gr" target=3D"_blank">kostas@physics.auth.gr</a>&gt=
;</span> wrote:<br>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">Hello,<br>
<br>
Recently an aging AFS cell we had online for several years, started experie=
ncing problems: ghost volumes not willing to be removed from vldb, volumes =
appearing with the same id but could be accessed by different names... we t=
raced it down to a corrupted vldb file, we fixed it using vldb_check and it=
 appears to behave again.<br>

<br>
While dealing with it, I also did a prdb_check and it came back with this:<=
br>
<br>
[root@srv-10-206-123-246 logs]# prdb_check -database prdb.DB0<br>
Bad user/group dicotomy in membership list<br>
=A0 Entry at 722624: flags 0x2, id -269i, next 722816.<br>
=A0 c:04/16/2006 22:02:02 a:06/28/2013 10:45:56 r:05/16/2013 14:38:43 n:tim=
e-not-set<br>
=A0 ids =A010207 =A010718 =A010772 =A011072 =A011120 =A012059 =A012134 =A01=
3108 =A013226 =A013247<br>
=A0 hash (id 0 name 136256). =A0Owner 1i, creator 1i<br>
=A0 quota groups 0, foreign users 0. =A0Mem: 57, inst: 0<br>
=A0 Owned chain 1155392, next owned 715520, inst ptrs(0 0 0).<br>
=A0 Name is &#39;webmasters&#39;<br>
=A0 =A0 Entry at 722816: flags 0x4, id -269i, next 1089728.<br>
=A0 =A0 c:time-not-set =A0 a:time-not-set =A0 r:time-not-set =A0 n:time-not=
-set<br>
=A0 =A0 ids =A013598 =A010611 =A011608 =A011137 =A012117 =A010027 =A011584 =
=A014027 =A012365 =A0 -310<br>
=A0 =A0 =A0 =A0 =A011123 =A012031 =A014013 =A011127 =A013010 =A014024 =A013=
241 =A012050 =A013728 =A013202<br>
=A0 =A0 =A0 =A0 =A010927 =A014768 =A010001 =A013381 =A012478 =A014283 =A011=
944 =A012970 =A013323 =A015396<br>
=A0 =A0 =A0 =A0 =A013197 =A010086 =A014770 =A011444 =A015628 =A011676 =A013=
216 =A014263 =A012686<br>
Entry membership count is inconsistent: 56 entries refer to this one<br>
=A0 Entry at 722624: flags 0x2, id -269i, next 722816.<br>
=A0 c:04/16/2006 22:02:02 a:06/28/2013 10:45:56 r:05/16/2013 14:38:43 n:tim=
e-not-set<br>
=A0 ids =A010207 =A010718 =A010772 =A011072 =A011120 =A012059 =A012134 =A01=
3108 =A013226 =A013247<br>
=A0 hash (id 0 name 136256). =A0Owner 1i, creator 1i<br>
=A0 quota groups 0, foreign users 0. =A0Mem: 57, inst: 0<br>
=A0 Owned chain 1155392, next owned 715520, inst ptrs(0 0 0).<br>
=A0 Name is &#39;webmasters&#39;<br>
=A0 =A0 Entry at 722816: flags 0x4, id -269i, next 1089728.<br>
=A0 =A0 c:time-not-set =A0 a:time-not-set =A0 r:time-not-set =A0 n:time-not=
-set<br>
=A0 =A0 ids =A013598 =A010611 =A011608 =A011137 =A012117 =A010027 =A011584 =
=A014027 =A012365 =A0 -310<br>
=A0 =A0 =A0 =A0 =A011123 =A012031 =A014013 =A011127 =A013010 =A014024 =A013=
241 =A012050 =A013728 =A013202<br>
=A0 =A0 =A0 =A0 =A010927 =A014768 =A010001 =A013381 =A012478 =A014283 =A011=
944 =A012970 =A013323 =A015396<br>
=A0 =A0 =A0 =A0 =A013197 =A010086 =A014770 =A011444 =A015628 =A011676 =A013=
216 =A014263 =A012686<br>
=A0 =A0 Entry at 1089728: flags 0x4, id -269i, next 0.<br>
=A0 =A0 c:time-not-set =A0 a:time-not-set =A0 r:time-not-set =A0 n:time-not=
-set<br>
=A0 =A0 ids =A012841 =A014278 =A013208 =A015078 =A011212 =A015651 =A013259 =
=A012842<br>
Entry membership count is inconsistent: 2 entries refer to this one<br>
=A0 Entry at 844544: flags 0x2, id -310i, next 0.<br>
=A0 c:12/19/2007 15:14:51 a:12/19/2007 15:52:37 r:time-not-set n:time-not-s=
et<br>
=A0 ids =A011123<br>
=A0 hash (id 0 name 418688). =A0Owner -269i, creator 1i<br>
=A0 quota groups 0, foreign users 0. =A0Mem: 1, inst: 1<br>
=A0 Owned chain 0, next owned 795200, inst ptrs(0 -269 0).<br>
=A0 Name is &#39;webmasters:<a href=3D"http://alumni.physics.auth.gr" targe=
t=3D"_blank">alumni.physics.<u></u>auth.gr</a>&#39;<br>
<br>
The cell has 4 prdb/vldb servers, 3 of them are version 1.4.5 and the last =
one, 1.6.6.<br>
<br>
The newer versioned server was installed just recently because the vldb_che=
ck from 1.4.5 could not fix the vldb errors. Along the same lines, prdb_che=
ck from 1.4.5 distribution does not report any errors on prdb, while the on=
e from 1.6.6 gives the above mentioned errors.<br>

<br>
Though not alarming to my eyes at least and no evident problem on the cell =
so far, we would like to fix the prdb problems as well. Question is how? pr=
db_check utility is missing a -fix flag.<br>
<br>
<br>
Thanks for any input,<br>
<br>
-Kostas.<br>
<br>
<br>
</blockquote></div><br><br clear=3D"all"><br>-- <br><div dir=3D"ltr">D</div=
>
</div>

--001a1133352ef53aa604f31e95eb--