[Date Prev][Date Next]
[Chronological]
[Thread]
[Top]
Re: Why built-in schemas are a bad idea
- To: Ulrich Windl <Ulrich.Windl@rz.uni-regensburg.de>, openldap-technical@openldap.org
- Subject: Re: Why built-in schemas are a bad idea
- From: Michael Ströder <michael@stroeder.com>
- Date: Wed, 7 Nov 2018 18:29:35 +0100
- Autocrypt: addr=michael@stroeder.com; prefer-encrypt=mutual; keydata= xsBNBFbdnRoBCADj0vYA4aRwKJ6AE4mf8oElLgMT/1eLNKpJ2FYBWcwj9d8dTk5/p9b8DRxy S/qQIUUZqt9xRFZwUCm0vFeQMRDeN9xzAKoRzrJifoDOacOjG1lhZTKYvVZGgUT89Ao3QeHh Q7gPzcAKNoueoR2y3FXStOYuRrbk5PlSjVAITjsotgc7PWE9mmVYpeu8a+byK/DBHKUyolOA 1UXYvDa7MbPhMtdNm8qnwtKs1Vsyk1VkErM+5cIe+zTT6WYQcmZMRjCtWGiFTzk9W6Mdlskk WRTKhKNgokTsgcy1ecaCBUZWxv/SyXgD81+rwRi9b8Px+1reg43ayxi8sV7jrI1feybbABEB AAHNJ01pY2hhZWwgU3Ryw7ZkZXIgPG1pY2hhZWxAc3Ryb2VkZXIuY29tPsLAdwQTAQgAIQUC Vt2dGgIbAwULCQgHAgYVCAkKCwIEFgIDAQIeAQIXgAAKCRAH3HrjaovJOFpTCACjO773gcmJ KvzjiNpUFl/gANyaJgIq4VbMQ7VthRb1F9X6YbdJ6Z99ntyESjGFCpjofcSomr2vJDpv6ht+ lY33yo20YwsMpqe2OeId0jPybG+FtabKjgBNoAk7iqnBGUvE4t0dz0n1LQVCQR2jxyTKmcNq OYpsRZ3H+6kWwJMuVgsNZglINVZ8JgV5QuLYN5jhYz+pOuFnU11bV6nWREvzZXzebe7g7Zus 6AsWjtJ0lDvgBNzLlF3/eFrVch6Bejs0SvuFseIdZQk+4YU6Rb8xul/jDFXIfo7eTmijO3dV T5AmC1cUi8czncwpgAJnEH8vYv23RoN/aw2gSMCS2huIzsBNBFbdnRoBCAC7L1cTVBVZZuM/ yxSUM5CsgGBlTD1Cr7C2ngZFsHSYXVLq6NUB8GZA2iLK96CrwnFw4/Jjz4llOjc50iVRMQKL RyFWOJAMrpPq2ew5T+Uoo524D//dwVbqkFVVuvM8NPiKIDyPGCjP+acM1D8hXwhOXgQ8Iz8Q 3/GRSYjitn9JrkF0ia2nhariznBKVu0LDffxF/hOCx45+QRR2/rYYlshfZMB7nEJX9P+hVfM CSzltz9Z8CldeUbiJvnyrISReR2XBw9oh8JkIUP0BtpIaify9A7EfzOk+W9BUnWe+YwdSUsB fJxOhSv+umyW5GMqZGFu+4oYnkzbe+1LUs1JarCtABEBAAHCwF8EGAEIAAkFAlbdnRoCGwwA CgkQB9x642qLyTjEUgf+JX6Atatl/QKe37yCj1OZYNPd3B0rPLJRF5mEmrADRXLZC9+uFeDS Wxxln040gnR6rjBHrRcvVmlTDiZY26iuL16+V+0/aZ9uyXNQSzk2cwDSiI/8gvr72Y+FN5fh cGXpeNHxHilYc9onzDhxyE76cwzqTKm4q2ULIH2u9IHQ5O86Fv6nHPYhe2fy1bhQapNwi/Xl 3G3i2WNH/w7m+1zWU1IddZOjmXzoxLT1BATwXGa0Tt5RjVb2mM1Wg3Zj6kqFkF2vvKcvrwj0 q0Ap5uyfN5m0uWzQMCMoaV9HQf7f5MkS1lnwBqDgnojjVAieX5uk7olUiRuPKHMfhvXulYP8 AA==
- In-reply-to: <5BE2B8ED020000A10002DFB3@gwsmtp1.uni-regensburg.de>
- Openpgp: id=43C8730E84A20E560722806C07DC7AE36A8BC938
- References: <5BE2B8ED020000A10002DFB3@gwsmtp1.uni-regensburg.de>
- User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:60.0) Gecko/20100101 Thunderbird/60.3.0
Ulrich,
while I somewhat understand your frustration...
On 11/7/18 11:05 AM, Ulrich Windl wrote:
> This makes a replicated cn=config very difficult for a rolling upgrade of MMR servers.
...I think the above is the root cause of your trouble.
Even if the build configuration would not have changed there has been
syncrepl interop issues between different releases in the past. So I
would not expect this to always work.
Personally I'd strongly recommend to use a decent config management
instead of replicating cn=config which helps with lots of other issues
too. Still rolling upgrades can be a challenge, but you would at least
not have to get cn=config working to get changes propagated.
(No, I don't want to start yet another static vs. dynamic config
discussion.)
Ciao, Michael.
Attachment:
smime.p7s
Description: S/MIME Cryptographic Signature