new python toys

My organization has hired a new programmer who's into Python and I've been learning from him. I was always able to use Python to get done whatever I needed to get done but there are plenty of things I didn't know about until he showed me. Here's a quick list:

• the * and ** operators
• Generators, though I have not yet used them (ideas)
• ipython: an interactive python shell (not necessarily IronPython). It's been allowing me explore the language quickly (much quicker to type '?' than google the documentation) and interactively; e.g.

  sh-4.2$ ipython
  Python 2.7.2 (default, Oct 27 2011, 01:40:22) 
  Type "copyright", "credits" or "license" for more information.
  
  IPython 0.12 -- An enhanced Interactive Python.
  ?         -> Introduction and overview of IPython's features.
  %quickref -> Quick reference.
  help      -> Python's own help system.
  object?   -> Details about 'object', use 'object??' for extra details.
  
  In [1]: 1+
    File "", line 1
      1+
        ^
  SyntaxError: invalid syntax
  
  
  In [2]: import ldap.<tab>
  ldap.async       ldap.dn          ldap.ldapobject  ldap.sasl        ldap.thread      
  ldap.cidict      ldap.filter      ldap.modlist     ldap.schema      ldap.threading   
  ldap.controls    ldap.functions   ldap.resiter     ldap.sys         ldap.traceback   
  
  In [2]: import ldap.filter
  
  In [3]: ldap.filter.<tab>
  ldap.filter.escape_filter_chars  ldap.filter.filter_format        
  
  In [3]: ldap.filter.escape_filter_chars?
  Type:       function
  Base Class: 
  String Form:
  Namespace:  Interactive
  File:       /usr/lib64/python2.7/site-packages/ldap/filter.py
  Definition: ldap.filter.escape_filter_chars(assertion_value, escape_mode=0)
  Docstring:
  Replace all special characters found in assertion_value
  by quoted notation.
  
  escape_mode
      If 0 only special chars mentioned in RFC 2254 are escaped.
      If 1 all NON-ASCII chars are escaped.
      If 2 all chars are escaped.
  
  In [4]:
  
  In [4]: ldap.filter.escape_filter_chars??
  Type:       function
  Base Class: 
  String Form:
  Namespace:  Interactive
  File:       /usr/lib64/python2.7/site-packages/ldap/filter.py
  Definition: ldap.filter.escape_filter_chars(assertion_value, escape_mode=0)
  Source:
  def escape_filter_chars(assertion_value,escape_mode=0):
    """
    Replace all special characters found in assertion_value
    by quoted notation.
    
    escape_mode
        If 0 only special chars mentioned in RFC 2254 are escaped.
        If 1 all NON-ASCII chars are escaped.
        If 2 all chars are escaped.
    """
    if escape_mode:
      r = []
      if escape_mode==1:
        for c in assertion_value:
          if c < '0' or c > 'z' or c in "\\*()":
            c = "\\%02x" % ord(c)
          r.append(c)
      elif escape_mode==2:
        for c in assertion_value:
          r.append("\\%02x" % ord(c))
      else:
        raise ValueError('escape_mode must be 0, 1 or 2.')
      s = ''.join(r)
    else:
      s = assertion_value.replace('\\', r'\5c')
      s = s.replace(r'*', r'\2a')
      s = s.replace(r'(', r'\28')
      s = s.replace(r')', r'\29')
      s = s.replace('\x00', r'\00')
    return s
  
  In [5]:
  ...
  In [60]: import string
  
  In [61]: string.<tab>
  string.Formatter        string.capwords         string.ljust            string.rsplit
  string.Template         string.center           string.lower            string.rstrip
  string.ascii_letters    string.count            string.lowercase        string.split
  string.ascii_lowercase  string.digits           string.lstrip           string.splitfields
  string.ascii_uppercase  string.expandtabs       string.maketrans        string.strip
  string.atof             string.find             string.octdigits        string.swapcase
  string.atof_error       string.hexdigits        string.printable        string.translate
  string.atoi             string.index            string.punctuation      string.upper
  string.atoi_error       string.index_error      string.replace          string.uppercase
  string.atol             string.join             string.rfind            string.whitespace
  string.atol_error       string.joinfields       string.rindex           string.zfill
  string.capitalize       string.letters          string.rjust            
  
  In [61]: string.whitespace
  Out[61]: '\t\n\x0b\x0c\r '
  
  In [62]: string.punctuation
  Out[62]: '!"#$%&\'()*+,-./:;<=>?@[\\]^_`{|}~'
  
  In [63]:
  

Zimbra kills killer feature in favor of "just use vmware"

Long have I looked forward to Zimbra having disaster recovery through server to server sync. It would have made things better in my organization in a DR situation. In lieu of not having this feature we live mirror replicate the Zimbra store server LUNs between data centrers on our XIVs. So our DR strategy is to use our backup store servers on the replicated LUNs. Setting this up would take some time. It would be much better if they were already in the Zimbra cluster all the time and flipping them over to be the master would be a "supported feature" not something a sys admin has to do by configuring another instance of the store server by using the same data.

Zimbra promised me this feature years ago and today they backed out and ask that we just implement this feature by running Zimbra on VMware and buying vCenter Site Recovery Manager. However, we don't want proprietary lock-in reaching down into our virtualization platform so we don't use VMware and instead use KVM. We like Zimbra because it's built with non-proprietary components aside from the active-sync and Outlook connectors which we liked having since lots of people want an Exchange-like interface and we can keep the proprietary code/protocols in their own little bucket.

Zimbra is calling lots of attention to VMware's features as opposed to what a future version of Zimbra was supposed to have:

"This enhancement requests asks to sync an active server to a passive server, we'll never do this. It's a legacy failover model when their are new technology options to provide elastic scaling and (geo)redundancy with stateless application servers and distributed data. It's taken some time for these technologies to progress to the point where they are viable enough for applications like Zimbra that need data consistency and atomic transactions."

Let's look at this more simply. VMware can sync VM objects. Zimbra was supposed to be able to sync Zimbra objects. If you run Zimbra on VMware then you can just use the sync'd VMware objects for DR but if you don't run VMware then it would be nice to have sync'd Zimbra objects. Sycing objects to a recovery site is not legacy. What's changed is the type of objects.

Update: Zimbra re-opened the feature request. I'm glad the community put enough pressure on them.

git redaction

Redaction in git on stack over flow

iptables be liberal cheatsheet

A while ago I posted about how Cisco firewalls can make certain packets' state look invalid to iptables which causes slow performance because packets would be dropped and had to be retransmitted. This was inspired by a much more interesting post on endpoint.com. Since then I've often had to relax the invalid checks on iptables by setting the following proc values on rhel5 or rhel6:

rhel5
/proc/sys/net/ipv4/netfilter/ip_conntrack_tcp_be_liberal

rhel6:
/proc/sys/net/netfilter/nf_conntrack_tcp_be_liberal

emacs usb foot switch

Found an entry on emacswiki.org about foot switches. I see that xkeys now has a GPL'd SDK. I think it's time to get my feet in on the action...