| Commit message (Collapse) | Author | Age | Files | Lines |
... | |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: Allow DSMLGW to build with GAR build scripts
Reviewed by: nhosoi@redhat.com (Thanks!)
Fix Description: Add a new macro DSMLGWJARS_BUILD_DIR which defaults to
dist/classes for internal builds. For external builds, the developer
can grab these jars from their respective locations, or grab the bundle
from the fds download site, which is what the GAR builds will do. The
location is then passed in on the make command line as
DSMLGWJARS_BUILD_DIR=/path/to/dsmlgwjars. Regular internal builds
should continue to work as always.
Platforms tested: RHEL4
Flag Day: no
Doc impact: no
|
|
|
|
| |
Reporting the patch generation code to the trunk.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: Get rid of nsperl; use perldap with system perl
Reviewed by: Noriko, Rob, Nathan (Thanks!)
Branch: HEAD
Fix Description: All perl scripts are made executable by using the
#!/usr/bin/env perl *nix trick. This means that the correct version of
perl must be in the user's PATH e.g. 5.6.1 or later. This version is
either shipped with the OS or available on all platforms. On HP/ux, it
is available as a depot which is installed in /opt/perl. For CGI perl
scripts, the PATH can be set in the admserv.conf, so we may have to do
that for HP/ux. To make perldap work, some ugly hacks are involved.
Each perl script that uses perldap has a BEGIN section that figures out
where it is in the server root, sets a server root variable, and sets
LD_LIBRARY_PATH and SHLIB_PATH to point to serverroot/shared/lib.
Perldap will be installed under serverroot/lib/perl. This directory
will have 3 subdirectories: arch - containing the binary files; auto -
containing autoloaded perl modules; and Mozilla - containing the base
perldap .pm files. The BEGIN section also sets the perl INC path to
find those modules. The directory gets rid of nsperl plus a lot of old
crufty perl building code that we do not use anymore. Those are the
removed files. The admin server code also gets rid of the perl.c wrapper.
Noriko pointed out that this does not take care of upgrade install, so I
added several more files and diffs to take care of that case.
Basically, go through the tasks in o=netscaperoot and replace
perl?scriptname with just scriptname. Also, go through all of the
template generated scripts and replace the shebang line with
#!/usr/bin/env perl, and make sure they are chmod +x. I also found a
few more places that referenced nsperl and removed them.
Platforms tested: RHEL4
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
| |
improper use of ldap_get_next.
|
|
|
|
| |
Ported the patch making code from Directory71RtmBranch to the trunk.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: Build Cleanup - open source AS, other components;
remove Fortezza; etc.
Reviewed by: Noriko, Nathan, Rob C. (Thanks!)
Fix Description: This allows us to build DS entirely outside of the
firewall with entirely open source components, including setuputil,
adminutil, adminserver, and java components. I still need to address
some issues around nsperl, perldap, dsmlgw, xmltools, and general ease
of build. This also gets rid of the crufty Fortezza build stuff and
addresses some other minor build issues.
Platforms tested: RHEL4
Flag Day: yes, but the internal builds should not be affected
Doc impact: wiki
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
|
|
|
| |
certain changes to not get synched.
|
|
|
|
| |
Ported the patch making code to the trunk.
|
| |
|
|
|
|
| |
dependency for java-1.4.2-ibm package.
|
| |
|
|
|
|
|
|
|
| |
To fix the bug "[160003] db2index.pl cannot find libldap50.so if only certain
parameters are used", template-db2index.pl has been modified.
When upgrading existing servers, the perl scripts are not touched unless forced
to. This upgradeServer provides the framework.
|
| |
|
|
|
|
| |
if the instance dir exists, run ns-config with "-r".
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: Use all open source components
Reviewed by: Noriko (Thanks!)
Fix Description: This changes the DS 72 build to pick up the new Admin
Server open source candidate (using Apache, modules, etc), and the new
adminutil and setuputil components. The code has been changed to
reflect the new naming (adminsdk -> adminutil, setupsdk -> setuputil)
and new versions of these components (7.1) and new /s/b/c disk layout
where applicable. We don't really get very much from the admin server
anymore, just the admin and base packages. All of the other setup stuff
(setup, setup.inf, svrcore, etc.) come directly from the setuputil package.
Platforms tested: RHEL3
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
|
|
|
|
| |
1) Makefile: Added debrandDirectory target for debranding
2) nsdefs.mk: Introduced a new macro NOSP_DIR_VERSION.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: ./ns-slapd crashes on bind containing invalid dn and password
Reviewed by: Noriko (Thanks!)
Branch: HEAD
Fix Description: It's really crashing on the search request. The problem is that the server assumes all strings are encoded in utf8 format, since that is the only encoding allowed by the LDAP standards. Non-utf8 works in most places except the function slapi_utf8StrToLower(), which returns NULL given a string of non-utf8 bytes. The fix for this particular problem is to check for a NULL return value and handle accordingly.
The real solution to this problem would be for the server to check for valid utf8 strings in _all_ LDAP data, according to the syntax of the attribute (e.g. for binary or octet string syntax data, and other binary formats, all bets are off, but then we shouldn't be doing strtolower on these blobs either).
And, while we're at it, add data validation based on syntax for _all_ attributes e.g. in a pre-op.
Platforms tested: RHEL4
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: slapd crashes during SASL authentication
Reviewed by: Noriko (Thanks!)
Branch: HEAD
Fix Description: I could not reproduce the crash. I tried several different ways - no password in entry, empty password in entry, SSHA hashed password in entry - no crashes. No useful information from the FDS bug reporter either. In fact I found that SASL Digest-MD5 was not working at all. We needed to use the SASL_AUX_PASSWORD_PROP define instead of hardcoding "userpassword" - I guess sasl is case sensitive. I also fixed some missing new lines in log messages.
Platforms tested: RHEL4
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: new instance creation creates error aci (2 types in RDN)
Reviewed by: Nathan (Thanks!)
Fix Description: This only seems to occur when logging in to the console as a user other than the Console Admin user (e.g. as Directory Manager in my tests). We need the Console Admin DN or user id to construct the ACIs. This value is held in the suitespot3x_uid form parameter. I had removed it while working on this bug or a related bug earlier, but now that I've added it back, everything seems to be working again.
Platforms tested: RHEL4
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
|
|
|
|
|
|
|
| |
used
In the perl script db2index.pl, before executing any ldap client command line
tools, should have chdir to the <dsroot>/shared/bin, where the rpath is set
from.
|
|
|
|
|
|
|
|
| |
1) Eliminated SLAPD_MODUTIL_TREE_THREASHHOLD from attr.c as well as valueset.c.
With this change, if an attribute has more than 1 value to add/replace/delete,
it creates an AVL tree to check the duplicates.
2) Replace was not checking the duplicated value at all. Added a code to put
the attribute values into the AVL tree as being done for add and delete.
|
|
|
|
|
|
| |
By default, it dumps the entire entry. (it used to be truncated at the BUFSIZ
size). In case no need to dump the entire entry, introduced the truncate option
"-t".
|
|
|
|
| |
access of memory which sometimes resulted in crashing.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: Daily Acceptance: Directory Install failed to register Directory server as a Red Hat server (81)
Reviewed by: Nathan (Thanks!)
Fix Description: The index code, in the replace case, was not checking to see if there were
actually any values to delete before attempting to delete them. This fix just
checks to see if there are any values to delete.
Platforms tested: RHEL3
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing with the new indexing tests
New Tests integrated into TET: none
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: Directory Server crashes when deleting a view
Reviewed by: Nathan (Thanks!)
Fix Description: Needed to pass in the _address_ of theCache.pCacheViews to views_cache_add_ll_entry. Yet another lesson in using the compiler to catch type errors rather than casting to void*.
Platforms tested: RHEL3
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: Adding multiple attributes using a single ldapmodify crashes ns-slapd
Reviewed by: Nathan (Thanks!)
Fix Description: In C, the array '[]' dereference operator takes precedence over the '*' deref operator. In this case, I needed to put parentheses around the pointer dereference to avoid having array dereferenced first. modary is a pointer to an array, not an array, so I can't dereference it with the array operator until I first dereference the pointer.
Platforms tested: RHEL3
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
|
|
|
| |
subtrees that the operation applies to
|
|
|
|
|
| |
in a critical section. This fix creates a mutex at init time and uses that
mutex to create a critical section around all PAM API access.
|
|
|
|
| |
returning SUCCESS.
|
|
|
|
|
|
|
| |
Coding done by David Irving, Fred Brittain, and Aaron Gagnon
Reviewed by Rich Megginson - minor changes to md5_pwd.c
Tested on RHEL3 with FDS post-7.1
Does not include the OpenLDAP migration script - that will be handled separately
|
|
|
|
| |
Plus fixed branding/version number.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: Reliab 05: chaining backend test core dumped on HPUX backend
Reviewed by: Noriko (Thanks!)
Fix Description: This is a 64 bit issue. For ber_scanf, the "i" and "e" flags require a pointer to a long. There were a couple of places in the chaining backend code where we were passing in an int instead. This works fine on 32 bit where ints and longs are both 32 bits. I did a quick check through the code - these were the only places not using longs. The only way you would see this bug is in a "hub" chaining backend - one that gets requests from a mux and chains them to another farm server, or possibly on a mux in certain conditions.
Platforms tested: HP-UX 11.i 64 bit
Flag Day: no
Doc impact: no
QA impact: rerun chaining tests
New Tests integrated into TET: none
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: Gateway adds 'ou' and 'l' attributes to NT groups
Reviewed by: David (Thanks!)
Fix Description: Remove the ou and l attributes from the NT group add, edit, and display forms.
Remove ou and l from the nt groups documentation.
Also removed nt group type. We don't need it anymore, and the default is "global" which is fine.
Platforms tested: RHEL4
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Bug Description: dsgw: online help for objectclass NTgroup
Reviewed by: Noriko (Thanks!)
Files: see diff
Branch: HEAD
Fix Description: I've removed all references to NT domains. Also, the docs say that
ntGroupDomainId has been deprecated in favor of ntUserDomainId.
Platforms tested: RHEL4
Flag Day: no
Doc impact: no
QA impact: should be covered by regular nightly and manual testing
New Tests integrated into TET: none
|
|
|
|
| |
match
|