HOWTO INFORMATIQUE

Cette page donne des réponses à diverses questions informatique

Voir aussi ComputerScience

WIKI

Voir aussi InformationAboutThisTwiki

Tutoriel : TWikiTutorial

Niveau Utilisateur

Créer un raccourci vers une icône :

Par exemple, créer comme raccourci de l'icône TODO :

Editer TWikiPreferences

* Set TD = http://planetowiki.irap.omp.eu/pub/TWiki/TWikiDocGraphics/todo.gif
* Set DO = 

Commentaire

Html comment = 
Twiki comment = # comment

Voir ce qui change sur le site

Pour savoir ce qui a changé (en ordre chronologique) sur le web Team, consulter cette page : WebChanges

Changements survenus sur le web "twiki" : WebChanges

on peut aussi s'inscrire sur cette page pour recevoir les changements du web Team par email : WebNotify

Contribuer (enrichir le site)

Vous êtes vivement encouragés à contribuer au contenu de ce site.

Pour modifier n'importe quelle page du site :

• cliquez sur bouton "Edit", faites vos modifs, puis cliquez sur le bouton "Save".
• si vous êtes plus à l'aise avec la souris, cliquez plutot sur le bouton "WYSIWYG"

Faire une copie (de sauvegarde) d'un Web (pour consultation offline) :

• Grace au plugin PublishContrib, on peut obtenir une copie du Web à tout instant (attention, cette copie est READ ONLY, ce n'est plus un wiki, mais juste des pages web).
• Pour publier la version courante du Web, aller sur le topic PublishWeb, et cliquer sur le bouton "Publish" (en milieu de page). Par défaut, c'est l'option "zip" qui est sélectionnée, ce qui crée un zip du Web.
• Il suffit maintenant de cliquer en bas de la page générée par la requete, sur le lien "Published To: NomduWeb.zip", pour télécharger ce zip (L'accès à cette adresse est protégé pour que seuls les personnes enregistrées puissent y acceder)
• Dézipper et cliquer sur la page WebHome.html. Le résultat n'est pas très joli, mais le contenu y est...
• Un historique de la dernière publication se trouve dans le topic PublishContribHistory (utile pour voir si tout s'est bien passé)

Mise à jour d'une pièce attachée (remarque) :

Il vaut mieux utiliser des noms de fichier assez génériques (par exemple FDD.pdf au lieu de FF_v1.4.pdf) ainsi on peut mettre à jour un document sans changer son nom et les liens vers ce fichier

Insérer une icone:

%Y%
%TD% ou %ICONURL{todo}%
%DO% ou %ICONURL{done}%
%CL% ou %ICONURL{closed}%

RACCOURCIS PRATIQUES

• ActionTrackerPlugin info : http://twiki.org/cgi-bin/view/Plugins/ActionTrackerPlugin et https://www.seegrid.csiro.au/twiki/bin/view/TWiki/ActionTrackerPlugin

• Liste de tous les users twiki enregistrés : UserList

• Liste plus générale des users twiki : TWikiUsers

• Liste de tous les groupes du twiki : TWikiGroups

• Liste de tous les topics : WebTopicList

• Notifications : WebNotify

Niveau Administrateur

Voir aussi InformationAboutThisTwiki

IDL

Conseils pour bien documenter son code

• Voir conseils généraux (Michael Galloy) "A style guide" : styleguide.pdf

• Voir IDLDoc (tutoriel) : idldoc-tutorial.pdf

• Voir IDLDoc (ref) : idldoc-reference.pdf

Je livre ici une synthèse des conseils de Michael Galloy pour bien documenter un code IDL (Lire "A style guide" pour la version complète avec justifications) :

1. General principles

1. Code is for humans.

2. Do not mix styles.

3. Avoid redundancy.

4. Use an easy to maintain style.

2. Layout

1. Layout should enhance the logical structure of the code.

2. Use two spaces (not tabs) per indentation level.

3. Use a maximum line length of 79 characters.

4. Write code in paragraphs.

Example:

function mg_sample, nValues, nIndices, seed=seed 
  compile_opt strictarr 

  ; get random nIndices by finding the indices of the smallest 
  ; nIndices in an array of random values 
  values = randomu(seed, nValues) 

  ; our random values are uniformly distributed, so ideally 
  ; the nIndices smallest values are in the first bin of the 
  ; below histogram 
  nBins = nValues / nIndices 
  h = histogram(values, nbins=nBins, reverse_indices=ri)

  ; the candidates for being in the first nIndices will live in 
  ; bins 0..bin 
  nCandidates = 0L 
  for bin = 0L, nBins - 1L do begin 
    nCandidates += h[bin] 
    if (nCandidates ge nIndices) then break 
  endfor 

  ; get the candidates and sort them 
  candidates = ri[ri[0] : ri[bin + 1L] - 1L] 
  sortedCandidates = sort(values[candidates]) 

  ; return the first nIndices of them 
  return, (candidates[sortedCandidates])[0:nIndices-1L] 
end 

5. Insert two blank lines between routines.

One blank line separates "paragraphs"; two blank lines separate "sections."

3. Comments

1. Document intent.

2. Keep comments up to date with the code.

3. For a short comment, use a phrase with lowercase first letter and no period. For longer comments, use all normal grammar rules.

5. Write a complete header for each routine.

Use comments between ;+ and ;- before the routine. Document the purpose of the routine, the return value (if a function), and side effects of the routine (which you should strive to eliminate). Each parameter should be documented with whether it is an input and/or output, optional or required, data type expected, default value (if any) and a description of its purpose.

For example, the below is an IDLdoc formatted header for the routine listed above:

;+ 
; Get nIndices random indices for an array of size nValues (do 
; not repeat an index). 
;
; :Returns: lonarr(nIndices) 
;
; :Params: 
;     nValues : in, required, type=long 
;          size of array to choose indices from 
;     nIndices : in, required, type=long 
;         number of indices needed 
;
; :Keywords: 
;     seed : in, out, optional, type=long or lonarr(36) 
;         seed to use for random number generation, a new seed 
;         will be output 
;- 
function mg_sample, nValues, nIndices, seed=seed 

6. Indent a comment along with the code it's documenting.

7. Document paragraph by paragraph.

Each paragraph of code may need a comment to document its purpose, but inside a paragraph use only end-of-line comments to comment particular lines. It can be helpful to write the comments first, providing an outline of the code to write.

ex of end-of-line comment :

if (event.type eq 2) then begin ; type 2 = motion events

9. Don't repeat the code in the comments.

Don't document the obvious.

10. Don't add extra comments for convoluted code; improve the code.

Don't document bad code—rewrite it. (Kernighan and Plauger, The Elements of Programming Style)

4. Statement formatting

1. Use lowercase for reserved words, operators, and compile_opt option names. For example, use: compile_opt strictarr if (not done) then readf, lun, line 2. Use only one statement per line. Multiple commands can be entered on one line using the & to separate them. This can be useful on the command line, but don't use it for code in files: routines, batch files, or main-level programs. For example, don't write: a = 1 & b = 2 3. Prefer begin/end blocks unless the entire statements fits on one line. For example, for a single short statement in a compound statement, use: for i = 0, 10 do print, i For a multiple statement block, write: for i = 0, 10 do begin j = i^2 print, j endfor For a single, long statement, use: for i = 0, 10 do begin print, i, format='("The index is ", I0, ".")' end But never write: for i = 0, 10 do $ print, i, format='("The index is ", I0, ".")' The same holds for each case of a case or switch statement:

Statement formatting 6 case uname of 'tlb': resize_widget, event.x, event.y 'draw': begin if (event.type ne 2) then return ; type 2 = motion events (*pstate).x = event.x (*pstate).y = event.y end else: endcase 4. Use specific end statements for the ending a block: endif, endwhile, endrep, endcase, and endswitch. Using the appropriate end statement lets the IDL compiler help catch structural errors in the code. There are no specific end statements for each case of a case or switch statement or the end of a routine or mainlevel program, so just use end. 5. Define structures one field per line unless the entire definition can fit on one line. For example, use: point = { x: 0.0, y:0.0 } state = { x: 0.0, $ y: 0.0, $ pdata: ptr_new(), $ ; image data drawId: 0L $ ; window identifier } Defining each field on its own line allows for individual fields to be found more easily and to be documented on the same line they are defined on. 6. Add one space around most operators. Normal arithmetic operators should have one space before and after them, for example: slope = (y0 - y1) / (x0 - x1) There are many exceptions to this rule. Don't put spaces around -> for method invocation or = for keyword use (but do for = for assignment). Don't put extra spaces around ()'s or []'s in any of their uses, but do add an extra space inside {}'s. Examples: tvscl, im, true=1 bin = arr[r[r[i]:r[i + 1] - 1]] point = { x: 0.0, y: 0.0 } Don't add extra spaces in order to align values. For example, do: x = 1 y = 2 longer = 3 instead of x = 1 y = 2 longer = 3 7. Use single quotes for strings. Both single and double quoted strings are allowed in IDL. Use only single quoted strings to avoid clashing with the strange octal value notation that uses double quotes,

Statement formatting 7 age = "22 years" is a syntax error because age = "22 defines age to be 22 in octal (18 in decimal). The additional years" in the previous statement is not understood by the IDL parser. Use double single quotes if you need a single quote. For example, sign = 'Eat at Joe''s' 8. Don't mix the bitwise operators and, or, and not with the logical operators &&, ||, and ~. Always use and, or, and not for all bitwise operations. For logical operations, either use the bitwise operators or (if using IDL 6.0 or later) use the logical operators &&, ||, and ~ along with the logical_predicate option to compile_opt. 9. Use capital letters to indicate the type of constant, use lowercase to indicate base of integers. Use 0L not 0l because "l" (lowercase letter el) looks like "1" (integer one). Use '5'o and '5'x, not '5'O and '5'X. Create constants of the correct type instead of creating a variable of the incorrect type and having IDL convert it. Though IDL will automatically convert it, that adds some overhead and, more importantly, obscures the type of the variable for readers of the code. 10. Use square brackets for array indices. Use compile_opt strictarr in each routine to prevent issues with IDL confusing arrays and functions calls. 11. Indent continuation lines of routine declarations and calls with the first argument. For example, do: pro mgitopadapthistequal::getProperty, clip=clip, $ nregions=nregions, $ top=top, $ _ref_extra=e 12. Indent continuation lines of assignment statements to the right of the =. For example, self.eyeEccentricity = n_elements(eye_eccentricity) eq 0 $ ? 0.5 $ : eye_eccentricity 13. Indent continuation lines of looping and conditional statements with the condition. For example, if (arg_present(sdev) $ || arg_present(variance) $ || arg_present(skewness) $

Variables 8 || arg_present(kurtosis)) then begin This can often be rewritten more clearly with the use of some variables that hold sections of the full expression. 14. Use labels sparingly for goto and ON_IOERROR statements. Use lowercase short names for labels. Indent labels with surrounding code. Place the label on its own line. For example, on_ioerror, io_problem openr, lun, filename, /get_lun arr = fltarr(100) readu, lun, arr free_lun, lun return, arr io_problem: print, err_string 5. Variables 1. Use good variable names. Think for a while. A thesaurus can be a valuable tool for finding the right name. Shorter names are acceptable if the variable is used only in a small section of code (such as a loop variable). Do use the standard names for common variables: state, pState, event, and tlb. 2. Variable names should be in camel case. Camel case uppercases the first letter of each word in the name except for the first letter. Uppercase each letter in an abbreviation that appears in a variable name unless it starts the name, i.e. noaaWeatherURL. Prefix a local variable with "my" when it mirrors a parameter, but has possibly been modified with a default value. Prefix the name with "n" for variables that hold counts, "o" for object references, "p" for pointers, "id" for iTools identifiers. Examples that follow this convention, myParam nFiles oModel pState idPlot 3. Prefer pointers, objects, and passing local variables over common blocks or system variables. Occasionally there is a reason to use common blocks or system variables, but you should have a good argument for it. When using direct graphics, prefer using graphics keywords of the plotting routines over setting system variables directly.

Routines 9 4. Define common blocks and named structures in only one location. Define the variables in a common block only once in a batch file. Include that file where needed. map_proj_init_commonblock.pro is an example of this. Define a named structure using automatic structure definition. For example, define MG_Point in a routine named MG_POINT__DEFINE in a file named mg_point__define.pro. 6. Routines 1. For any file containing IDL code, filenames should always be in lower case and use the ".pro" extension. Lowercase filenames reduce cross-platform issues. IDL will automatically find code that is in files using the ".pro" extension (provided it follows the other conventions listed in below). 2. Each file should include only one routine called from outside the file. Each file should contain only one routine called from outside that file. Add the ".pro" extension to the routine name of the externally called routine to get the filename. For example, the routine MG_LINEAR_FUNCTION should be in a file named mg_linear_function.pro. If there are multiple routines in the file, make sure the externally called routine is last in the file. Following this rule will insure that IDL automatically finds and compiles all code necessary (provided the files are in the IDL path or current directory). The names of the helper routines in the same file should be prefixed with the entire name of the external routine, such as MG_LINEAR_FUNCTION_HELPER. A prominent exception to this rule is the case of class definition files which should include all the methods of the class, many of which could be called from outside the file. IDL will still find the methods automatically if done following the style rules in the section on object-oriented programming. 3. Routine names should be lower case, begin with a short organization prefix, and separate words with underscores. The prefix indicates the individual or group responsible for the code. It is usually the initials of the individual or orgranization. Limit to two or three letters. Don't use the "IDL", "RSI", "ITT", "cw", "it", or empty prefixes. For example, here are some names following this style: mg_linear_function mg_itbrowser mg_sample 4. Routines should fit on one screen. It is much easier to understand a routine when it can viewed in its entirety. An exception are routines with case statements where there are many short cases. 5. Keywords should always be optional for the caller; positional parameters should generally be required. Keywords should either be an optional input with a reasonable default value or an extra output (i.e. not the main purpose of the routine). This allows new users of the routine to focus on the few positional parameters and examine the more numerous keywords as needed when the default values are not appropriate for them. 6. Keyword names should be lowercase and separate words with underscores.

Object-oriented programming 10 For example, filename ntests eye_separation left_image 7. Always use the full keyword name when calling the routine. The same abbreviations which are so handy when typing at the commandline become quite cryptic when written in a file and examined years later. They may even be ambiguous if new versions of the code they call have been released (with more keywords) in the meantime. Using the full name prevents ambiguity and increases the readability. 8. If the purpose of a routine is to return a value, use a function, otherwise use a procedure. If the main purpose of the routine is to perform some action (besides a query), but incidentally needs to return a value use a procedure with an output keyword. In general, use a function when it would be useful to chain together multiple calls in one expression. Avoid the C-style status as the function return value. 9. Status and error codes should be returned via keyword. Often this means that a function needs to return a null or impossible value if an error occurs during execution. For example, WHERE returns -1 when there are no matching elements of the given array. 10. Setup firewalls for error handling as appropriate to the application. Data that crosses a firewall should be checked carefully for correctness. There should always be a firewall around the entire program for data coming in (from files, network, or user). Other firewalls should be added as appropriate to the application: some may require a firewall around each routine, others between "packages", and others may require no other firewalls besides the outside wall. Use on_error, 2 in simple, short routines when you become confident that the only errors occurring in the routine are from bad inputs. 7. Object-oriented programming In addition to the other rules, there are a few extra guidelines for writing object-oriented code. 1. Class names should begin with a prefix indicating organization and a code indicating the class' area of use. Each word should be capitalized Use the same prefix as given to normal routine names; avoid "IDL", "ITT", "RSI", and the empty prefix. Codes already in use by IDL: an (analysis), com (COM), db (database), ex (example), ff (file format), gr (graphics), it (iTools), sys (system), net (network), and _ (general use). Make use of the existing codes and make up new ones as necessary.

Object-oriented programming 11 The examples, MGgrWindow3D MGffTemplate MGutTestCase use the codes gr and ff already in use by IDL, but creates a new code ut (for unit testing) because none is provided. Use all caps for abbreviations in class names, as in IDLnetURL. 2. Put all the methods and the routine defining the instance variables for a class into a single file. For the definition of MGexClass, the file should be named mgexclass__define.pro. The last routine in this file should be MGEXCLASS__DEFINE and should define the instance variables for the class (i.e. create a named structure with name MGexClass). Define only one structure/class name in the __DEFINE routine. 3. Method names should be a verb phrase in camel case. For example, here are some method names following these conventions: getProperty setProperty add findTestNames runTest reportTestResult Use the conventions that are used by the IDL library classes. For example, use the GETPROPERTY and SETPROPERTY scheme of procedures to handle getting and setting properties of a class. 4. Begin "protected" methods' names with a underscore. For example, MGexSomeClass::_helperMethod is a helper method called by other methods in MGexSomeClass, but should not be called from outside of MGexSomeClass. Because IDL has no mechanism for limiting who can call a method, the underscore merely indicates that only that class and its subclasses should call that method. This gives a visual cue to the caller without enforcing the protection. Outside callers can still call this method, but at least they have been warned. 5. Beware of multiple inheritance. Use multiple inheritance as a last resort. Prefer delegation for one of the parent classes i.e. make the new class contain the secondary parent class as an instance variable. Reference Listed below are some references that provided inspiration for the rules included in this style guide. [Custom99] Custom Visuals. “IDL Style Guide”. www.customvisuals.com/IDL_Style.html. 1999.

Object-oriented programming 12 [Fanning03] David Fanning. Coyote's Guide to IDL Programming. www.dfanning.com. “IDL Style Guide”. 2003. [ITTVIS07] ITT Visual Information Solutions. www.ittvis.com. “One Proposal for an IDL Coding Standard”. Tech Tip 4120. 2007. [KernPlauger78] Brian W. Kernighan and P. J. Plauger. The Elements of Programming Style. 2nd edition. 1978. [McConnell04] Steve McConnell. Code Complete. 2nd edition. 2004. [Rossum01] Python Enhancement Proposals. www.python.org/dev/peps. Guido van Rossum and Barry Warsaw. “Python PEP-8: Style Guide for Python”. 2001.

IdlDoc : générateur de documentation pour IDL

home: http://idldoc.idldev.com/

download: http://idldoc.idldev.com/wiki/Downloads

tutoriel: http://idldoc.idldev.com/wiki/GettingStarted

Installation: (NB: sur hyperion, c'est déjà installé sous /usr/local/itt/idldoc/idldoc)

1) Download

2) Copier et dézipper dans un dossier de mon choix (ex : D:/idldoc/idldoc-3.3)

3) Ajouter le chemin vers idldoc dans le IDL_PATH (Fenêtre/Preferences/IDL/Chemins/Insérer (puis Appliquer)

Utilisation:

Pour générer la doc avec idldoc :

IDL> idldoc, root='C:\Users\Etienne\IDLWorkspace80\monprojet\trunk\src', output='C:\Users\Etienne\IDLWorkspace80\monprojet\trunk\doc'

Puis, 2clic sur index.html

Enregistrer une image dans un fichier

; envoi vers fichier postscript
set_plot, 'ps'

; éventuellement donner un nom
device, filename='toto.ps'

plot, findgen(2)

; retour à l'affichage sur écran
device, /close
set_plot, 'x'

PHP

Afficher un bloc de code html dans php (ou tout autre texte long) :

http://forum.webrankinfo.com/petite-astuce-pour-afficher-simplement-code-html-t32812.html

LINUX

Niveau Utilisateur

SED (substitution de texte)

Pleins d'exemples ici : http://sed.sourceforge.net/sed1line_fr.html

• remplacer tutu par toto, mais seulement des lignes 3 à 6 :

cat monfichier.txt | sed -e 3,6"s/tutu/toto/" >| monfichiertemp.txt

• remplacer tutu par toto, mais seulement entre les lignes "titi" et "toto" :

/titi/,/toto/s/tutu/toto/

ou encore :

sed -e "/$debut/,/$fin/"'s#HostId="t3://.*\n#HostId="t3://'"$WLS_ADM_HOST:$PORT\"/" < "$file1" > "$file2"

GREP (recherche textuelle)

Afficher uniquement les lignes de commentaire:

egrep '^#' fichier

Ne pas afficher les lignes de commentaire:

egrep -v '^#' fichier

N'afficher que les lignes utiles (tout sauf commentaires et lignes vides) :

egrep -v '^(#|$)' fichier

RSYNC

Sauvegarder (de façon incrémentielle) des répertoires du poste A vers le poste B :

1) Depuis A...

A#> rsync -avz /rep/rep1 /rep/rep2 B:/rep/

option -v = verbose

attention aux slashes '/' :

/rep/rep1 écrit un répertoire rep1 sur B

/rep/rep1/ écrit directement le contenu du répertoire rep1 sur B

2) ... ou Depuis B

rsync -avz A:/rep/rep1 A:/rep/rep2 ./rep

FIND

Supprimer tous les répertoires CVS/ dans toute une arborescence (à partir du répertoire courant) :

find . -name "CVS" -exec \rm -r {} \; > /dev/null 2>&1

Combien de fichiers dans un répertoire (y-compris les sous-rep) :

find rep | wc -l

Affecter des droits spécifiques à certains éléments d'un répertoire (tout le contenu, y-compris sous-dossiers) :

dossiers:

find /home/jsmith/awstats/ -type d | xargs chmod 0755

fichiers:

find /home/jsmith/awstats/ -type f | xargs chmod 0644

fichiers perl:

find /home/jsmith/awstats/ -type f -name *.pl | xargs chmod 0755

Forwarder les mails du serveur

Par exemple, je veux forwarder mes mails de pallier@planetoweb.cesr.fr vers ma boite cesr ou encore ma boite perso :

cd ~

créer un fichier ".forward" contenant mon adresse cesr ou perso

S'il s'agit de forwarder les mails adressés à root@planetoweb.cesr.fr, on peut utiliser la même solution, mais on peut aussi ajouter cette ligne tout à la fin du fichier /etc/aliases

root: etienne.pallier@cesr.fr

Rechercher un fichier (dans une hiérarchie de répertoires)

find . -name filename_searched

Rechercher un mot dans un ou plusieurs fichiers

• rechercher "mot" dans les fichiers idl du répertoire courant

grep "mot" *.pro

• rechercher "mot" dans tous les fichiers du répertoire courant

grep -i "mot" .

L'option "-i" permet de ne pas tenir compte de la casse

• rechercher "mot" dans tous les fichiers du répertoire courant ainsi que dans tous les sous-rep (récursif) :

grep -r -i "mot" .

Faire une action sur un ensemble de dossiers et fichiers

Depuis répertoire courant, supprimer tous les dossiers "CVS" (récursivement) :

find . -name CVS -exec rm -rf {} \;

Créer un alias

Exemple : quand on tape "l" ça fera "ls -l"

alias l="ls -l"

Cette ligne doit être placée dans votre ~/.bashrc

Configurer mon environnement

Pour les configurations perso, les placer dans son ~/.bash_profile (ce script appelle ~/.bashrc qui doit contenir les alias et fonctions)

Mettre les configurations générales dans /etc/profile.d/profile_etienne.sh (elles seront ainsi valables pour TOUS les users du serveur)

Ce fichier est automatiquement lu par /etc/profile au démarrage d'une session

Imprimer sur linux

Pour installer l'imprimante "sprinter", voir SprinterPrinter

Pour faire de cette imprimante l'imprimante par défaut, ajouter cette ligne dans votre ~/.bash_profile :

export PRINTER=sprinter (mettre le nom de votre imprimante à la place de "sprinter")

Pour imprimer du texte, on peut utiliser les utilitaires "a2ps" ou "enscript"

Visualiser des images (jpeg, ps, ...) en mode console

Note : display permet aussi de convertir une image d'un format vers un autre (enregistrer sous...)

Voir aussi "xv" et "gthumb"

Convertir une image d'un format vers un autre

PS to PDF : ps2pdf

Voir aussi l'utilitaire convert

En mode interactif, On peut aussi utiliser display (voir ci-dessus)

Niveau Administrateur

LOGS

/var/log

auth.log ou secure (sur Redhat)

Outil de visualisation graphique des logs : gnome-system-log

logrotate

/etc/logrotate.conf

/etc/logrotate.d/

Forcer une rotation :

sudo logrotate -f /etc/logrotate.conf

==> new file "messages", "messages" devient "messages1", "messages1" devient "messages2"...

logwatch

by default, runs daily on yesterday's logs (/etc/cron.daily/0logwatch qui pointe sur /usr/share/logwatch/scripts/logwatch.pl), includes all services, and sends a mail to root

Configuration locale :

• Générale : /usr/share/logwatch/default.conf/logwatch.conf
• Locale : /etc/logwatch/conf/logwatch.conf

Exemples :

info sur l'activité sshd d'aujourd'hui :

logwatch --service sshd --range=Today

Plus de détails :

logwatch --service sshd --range=Today --detail=Medium (ou High)

Pour créer un fichier au lieu d'un mail : --save=logwatch.today

Bonne config de logwatch.conf :

Range=All (au lieu de "Yesterday") Detail=High Archives=Yes (ajoute les logs des semaines précédentes, pas seulement le log en cours)

Equivalent commande : --range=All --archives

Exemple de config sur planetoweb (aug 2010) :

Création d'une config locale /etc/logwatch/conf/logwatch.conf :

Range = All
Detail = High
Archives = Yes
Output = html
Save = /var/www/html/servers/planetoweb/logwatch.html

Résultat sur http://planetoweb/servers/planetoweb/logwatch.html

swatch

http://www.linux-mag.com/id/7807

yum install swatch

[root@planetoweb planetoweb]# rpm -ql swatch
/usr/bin/swatch
/usr/lib/perl5/vendor_perl/5.8.8/Swatch
/usr/lib/perl5/vendor_perl/5.8.8/Swatch/Actions.pm
/usr/lib/perl5/vendor_perl/5.8.8/Swatch/Threshold.pm
/usr/lib/perl5/vendor_perl/5.8.8/Swatch/Throttle.pm
/usr/lib/perl5/vendor_perl/5.8.8/auto/Swatch
/usr/lib/perl5/vendor_perl/5.8.8/auto/Swatch/Actions
/usr/lib/perl5/vendor_perl/5.8.8/auto/Swatch/Actions/autosplit.ix
/usr/share/doc/swatch-3.2.1
/usr/share/doc/swatch-3.2.1/CHANGES
/usr/share/doc/swatch-3.2.1/COPYING
/usr/share/doc/swatch-3.2.1/COPYRIGHT
/usr/share/doc/swatch-3.2.1/KNOWN_BUGS
/usr/share/doc/swatch-3.2.1/README
/usr/share/doc/swatch-3.2.1/examples
/usr/share/doc/swatch-3.2.1/examples/SendMail.pm
/usr/share/doc/swatch-3.2.1/tools
/usr/share/doc/swatch-3.2.1/tools/reswatch
/usr/share/doc/swatch-3.2.1/tools/swatch_oldrc2newrc
/usr/share/man/man1/swatch.1.gz
/usr/share/man/man3/Swatch::Actions.3pm.gz
/usr/share/man/man3/Swatch::Threshold.3pm.gz
/usr/share/man/man3/Swatch::Throttle.3pm.gz

/usr/share/doc/swatch

Créer un startup script : vi /etc/init.d/swatch

#!/bin/sh
# Simple Log Watcher Program

case "$1" in
'start')
		/usr/bin/swatch --daemon --config-file=/etc/swatch.conf --tail-file=/var/log/auth.log --pid-file=/var/run/swatch.pid
		;;
'stop')
		PID=`cat /var/run/swatch.pid`
		kill $PID
		;;
*)
		echo "Usage: $0 { start | stop }"
		;;
esac
exit 0

chmod 755 /etc/init.d/swatch

Make sure swatch starts automatically at my runlevels :

# ln -s /etc/init.d/swatch /etc/rc2.d/S99swatch
# ln -s /etc/init.d/swatch /etc/rc3.d/S99swatch
# ln -s /etc/init.d/swatch /etc/rc5.d/S99swatch

Créer un fichier de conf /etc/swatch.conf :

watchfor /invalid|repeated|incomplete/
         echo
	 write khess
	 mail addresses=khess@localhost, subject=Authentication Problems

/etc/init.d/swatch start ==> créer un pid dans /var/run/swatch.pid et un fichier /root/.swatch_script.xxxx (avec xxxx = pid - 2)

SSH, config plus stricte

vi /etc/ssh/sshd_config

Port 22 : on pourrait mettre ici un autre port, genre 2010, histoire de brouiller les pirates...

PermitRootLogin? no

LoginGraceTime? 2m 
changed to : 
LoginGraceTime? 30

/etc/init.d/sshd restart 

Surveiller un serveur

last reboot

lastlog

dmesg

lspci

/var/log/messages

/var/log/boot.log

Voir aussi section suivante (test des disques)

Tester les disques durs

Occupation disques : df -h

[root@hyperion home]# df -h
Sys. de fich.         Tail. Occ. Disp. %Occ. Monté sur
/dev/sda3             178G   63G  106G  38% /
/dev/sdb1             2,7T  1,4T  1,3T  52% /data
/dev/sda1              99M   29M   65M  31% /boot
tmpfs                  24G  4,0K   24G   1% /dev/shm

cat /proc/partitions

cat /etc/mtab

cat /proc/mounts (equivalent de la commande "mount")

cat /proc/scsi/scsi

Voir aussi si messages d'erreur disques dans /var/log/messsages

Tester aussi la commande dmesg (equivalent à /var/log/dmesg)

Yum

Annuler l'information sur les updates disponibles :

/etc/init.d/yum-updatesd stop

su -c 'chkconfig --level 2345 yum-updatesd off'

Every other workaround seems to only kill yum-updatesd for runlevel 5.

Try yum update after that.

Keep in mind that you will no longer be informed that there updates available and as such you will need to check periodically with yum check-update

Quels sont les packages commençant par "mesa-" installés + à installer ? --> yum list mesa-*

Samba

Créer un new user "toto" : sbmpasswd -a toto

Mettre à jour le pass de "toto" : smbpasswd toto

Est-ce que samba est à l'écoute ? : service smb status

Redémarrer samba : service smb restart

Depuis le poste client windows :

clic droit sur Poste de Travail / Créer un lecteur réseau / "\\nom_du_serveur\mon_nom"

Corriger un bug sur Mac Leopard (pas sur Tiger) : samba ne suit pas les liens qui vont vers un disque différent

Pour corriger ce bug, il suffit d'ajouter cette ligne dans la partie "[global]" du smb.conf :

unix extensions = no

WINDOWS

Migrer de Windows XP vers Windows 7

Il faut acheter la licence Microsoft. La démarche dépend de la marque du PC.

Pour un hp : aller sur le site hp et rechercher "windows upgrade"

Modèles éligibles : http://h41112.www4.hp.com/promo/win7web/fr/fr/eligmodels.html

http://welcome.hp.com/country/fr/fr/mda/windows7/upgrade/which_version.html?jumpid=reg_R1002_FRFR

http://h41112.www4.hp.com/promo/win7web/fr/fr/

Envoyer des mails avec Outlook quand on n'est pas au CESR

1) "Creuser" un tunnel avec Putty

Créer une nouvelle session que vous appelerez par exemple "tunnel_mail_cesr", et qui créera un tunnel entre le port local 9025 et le port SMTP (25) du serveur de mail cesr (fw-in.cesr.fr).

La procédure est décrite sur la faq informatique du cesr : http://www1.cesr.fr/intranet/informatique/faq/ssh-tunnels/ssh.htm#_Toc150503764 (Exemple 1, cas numéro 2)

Il faudra lancer cette session (et la garder ouverte en tâche de fond) avant d'envoyer un mail

2) Configurer votre client mail Outlook

• Menu Outils/Paramètres du compte
• Clic sur bouton "Modifier"
• Dans le champ "Serveur de courrier sortant (SMTP)", remplacer "mailhost.cesr.fr" par "localhost"
• Clic sur bouton "Paramètres supplémentaires...", clic sur onglet "Options avancées"
• Dans le champ "Serveur Sortant (SMTP)", remplacer "25" par "9025"

Bien sûr, il faut tout remettre en place quand on revient au cesr (en fait, on pourrait aussi fonctionner en permanence avec le tunnel, même au CESR...)

Synchroniser Outlook 2007 Calendar et Google Calendar

http://content.techrepublic.com.com/2346-10877_11-191016.html

Installer un serveur X sur Windows pour afficher les fenêtres graphiques de linux

On propose d'utiliser Xming qui est un serveur X gratuit pour Windows

Il suffit d'éxécuter Xming avant de lancer une session graphique via putty (ou autre shell)

Installation : http://sourceforge.net/projects/xming

Cliquer sur Download

1) Installer Xming

2) Installer Xming-fonts (surtout nécessaire pour Emacs)

-- EtiennePallier - 25 Dec 2009